This action might not be possible to undo. Are you sure you want to continue?

# 8-1

1234567898

8.1

a) For step response,

input is M t u · ′ ) ( ,

s

M

s U · ′ ) (

) (

1

1

) ( s U

s

s

K s Y

D

D

c a

′

1

]

1

¸

+ ατ

+ τ

· ′ =

1

]

1

¸

+ ατ

+ τ

) 1 (

1

s s

s

M K

D

D

c

· ′ ) (s Y

a

) 1 ( 1 + ατ

+

+ ατ

τ

s s

M K

s

M K

D

c

D

D c

Taking inverse Laplace transform

) 1 ( ) (

) /( ) /(

D D

t

c

t c

a

e M K e

M K

t y

ατ − ατ −

− +

α

· ′

As α →0

M K dt

e

t M K t y

c

t

t

c a

D

+

α

δ · ′

∫

∞

·

ατ −

0

) /(

) ( ) (

M K t M K t y

c D c a

+ τ δ · ′ ) ( ) (

Ideal response,

) ( ) ( ) ( s U s G s Y

i i

′ · ′ =

1

]

1

¸

+ τ

s

s

M K

D

c

1

= K

c

Mτ

D

+

s

M K

c

M K t M K t y

c D c i

+ δ τ · ′ ) ( ) (

Hence ) ( ) ( t y t y

i a

′ → ′ as 0 → α

For ramp response,

input is Mt t u · ′ ) ( ,

2

) (

s

M

s U · ′

Solution Manual for Process Dynamics and Control, 2

nd

edition,

Copyright © 2004 by Dale E. Seborg, Thomas F. Edgar and Duncan A. Mellichamp.

8-2

) (

1

1

) ( s U

s

s

K s Y

D

D

c a

′

1

]

1

¸

+ ατ

+ τ

· ′ =

1

]

1

¸

+ ατ

+ τ

) 1 (

1

2

s s

s

M K

D

D

c

· ′ ) (s Y

a

) 1 ( ) 1 (

2

+ ατ

+

+ ατ

τ

s s

M K

s s

M K

D

c

D

D c

1

]

1

¸

+ ατ

ατ

+ +

ατ −

+

1

]

1

¸

+ ατ

ατ

− τ ·

1

) ( 1

1

1

2

2

s s s

M K

s s

M K

D

D D

c

D

D

D c

Taking inverse Laplace transform

[ ] [ ] ) 1 ( 1 ) (

) /( ) /(

− ατ + + − τ · ′

ατ − ατ −

D D

t

D c

t

D c a

e t M K e M K t y

As α →0

Mt K M K t y

c D c a

+ τ · ′ ) (

Ideal response,

· ′ ) (s Y

i

1

]

1

¸

+ τ

2

1

s

s

M K

D

c

=

2

s

M K

s

M K

c D c

+

τ

Mt K M K t y

c D c i

+ · ′ τ ) (

Hence ) ( ) ( t y t y

i a

′ → ′ as 0 → α

b) It may be difficult to obtain an accurate estimate of the derivative for use

in the ideal transfer function.

c) Yes. The ideal transfer function amplifies the noise in the measurement by

taking its derivative. The approximate transfer function reduces this

amplification by filtering the measurement.

8.2

a)

1 1 ) (

) (

1

2 1 2 1

2

1

1

+ τ

+ τ +

· +

+ τ

·

′

s

K s K K

K

s

K

s E

s P

1

1

1

1

]

1

¸

+ τ

+

+

τ

+ ·

1

1

) (

1

2 1

1 2

2 1

s

s

K K

K

K K

8-3

b) K

c

= K

1

+ K

2

→ K

2

= K

c

−

K

1

D

ατ · τ

1

2 1

2

2 1

1 2

K K

K

K K

K

D

D

+

ατ

·

+

τ

· τ

or

2 1

2

1

K K

K

+

α

·

α · +

2 2 1

K K K

) 1 (

2 2 2 1

− α · − α · K K K K

Substituting,

1 1 1

) 1 ( ) 1 ( ) 1 )( ( K K K K K

c c

− α − − α · − α − ·

Then,

c

K K

,

_

¸

¸

α

− α

·

1

1

c) If K

c

= 3 , τ

D

= 2 , α = 0.1 then,

27 3

1 . 0

9 . 0

1

− · ×

−

· K

30 ) 27 ( 3

2

· − − · K

τ

1

= 0.1 × 2 = 0.2

Hence

K

1

+ K

2

= -27 + 30 = 3

2

3

2 . 0 30

2 1

1 2

·

×

·

+

τ

K K

K

,

_

¸

¸

+

+

·

1 2 . 0

1 2

3 ) (

s

s

s G

c

8-4

8.3

a) From Eq. 8-14, the parallel form of the PID controller is :

1

]

1

¸

τ′ +

τ′

+ ′ · s

s

K s G

D

I

c i

1

1 ) (

From Eq. 8-15, for α→0, the series form of the PID controller is:

[ ] 1

1

1 ) ( + τ

1

]

1

¸

τ

+ · s

s

K s G

D

I

c a

1

]

1

¸

τ +

τ

+

τ

τ

+ · s

s

K

D

I I

D

c

1

1

1

1

1

1

1

]

1

¸

,

_

¸

¸

τ

τ

+

τ

+

τ

,

_

¸

¸

τ

τ

+

+

,

_

¸

¸

τ

τ

+ ·

I

D

D

I

I

D I

D

c

s

s

K

1 1

1

1 1

Comparing G

a

(s) with G

i

(s)

,

_

¸

¸

τ

τ

+ · ′

I

D

c c

K K 1

,

_

¸

¸

τ

τ

+ τ · τ′

I

D

I I

1

I

D

D

D

τ

τ

+

τ

· τ′

1

b) Since

,

_

¸

¸

+

I

D

τ

τ

1 ≥ 1 for all τ

D

, τ

I

, therefore

c c

K K ′ ≤ ,

I I

τ′ ≤ τ and

D D

τ′ ≥ τ

c) For K

c

= 4, τ

I

=10 min , τ

D

=2 min

8 . 4 · ′

c

K , min 12 · τ′

I

, min 67 . 1 · τ′

D

d) Considering only first-order effects, a non-zero α will dampen all

responses, making them slower.

8-5

8.4

Note that parts a), d), and e) require material from Chapter 9 to work.

a) System I (air-to-open valve) : K

v

is positive.

System II (air-to-close valve) : K

v

is negative.

b) System I : Flowrate too high → need to close valve →decrease controller

output →reverse acting

System II: Flow rate too high →need to close valve → increase

controller output → direct acting.

c) System I : K

c

is positive

System II : K

c

is negative

d) K

c

K

v

K

p

K

m

System I : + + + +

System II : − − + +

K

c

and K

v

must have same signs

e) Any negative gain must have a counterpart that "cancels" its effect. Thus,

the rule:

# of negative gains to have negative feedback = 0 , 2 or 4.

# of negative gains to have positive feedback = 1 or 3.

8.5

a) From Eqs. 8-1 and 8-2,

[ ] ) ( ) ( ) ( t y t y K p t p

m sp c

− + · (1)

The liquid-level transmitter characteristic is

y

m

(t) = K

T

h(t) (2)

where h is the liquid level

K

T

> 0 is the gain of the direct acting transmitter.

8-6

The control-valve characteristic is

q(t) = K

v

p(t) (3)

where q is the manipulated flow rate

K

v

is the gain of the control valve.

From Eqs. 1, 2, and 3

[ ] [ ] ) ( ) ( ) ( ) ( t h K t y K K p t p K q t q

T sp c V v

− · − · −

) (

) (

t h K y

q t q

K K

T sp

c V

−

−

·

For inflow manipulation configuration, q(t)>q when y

sp

(t)>K

T

h(t). Hence

K

v

K

c

> 0

then for "air-to-open" valve (K

v

>0), K

c

>0 : reverse acting controller

and for "air-to-close" valve (K

v

<0), K

c

<0 : direct acting controller

For outflow manipulation configuration, K

v

K

c

<0

then for "air-to-open" valve, K

c

<0 : direct acting controller

and for "air-to-close" valve, K

c

>0 : reverse acting controller

b) See part(a) above

8.6

For PI control

,

_

¸

¸

τ

+ + ·

∫

t

I

c

dt t e t e K p t p

0

* *) (

1

) ( ) (

,

_

¸

¸

τ

+ · ′

∫

t

I

c

dt t e t e K t p

0

* *) (

1

) ( ) (

Since

e(t) = y

sp

– y

m

and y

m

= 2

8-7

Then

e(t)= -2

,

_

¸

¸

τ

− − ·

,

_

¸

¸

−

τ

+ − · ′

∫

t K dt K t p

I

c

t

I

c

2

2 * ) 2 (

1

2 ) (

0

Initial response = − 2 K

c

Slope of early response =

I

c

K

τ

−

2

− 2 K

c

= 6 → K

c

= -3

I

c

K

τ

−

2

= 1.2 min

-1

→ τ

I

= 5 min

8.7

a) To include a process noise filter within a PI controller, it would be placed

in the feedback path

b)

f

1

1 23 s

,

_

¸

¸

τ

+

s

K

I

c

1

1

c) The TF between controller output ) (s P′ and feedback signal Y

m

(s) would

be

8-8

) 1 (

) 1 (

) (

) (

+ τ τ

+ τ −

·

′

s s

s K

s Y

s P

f I

I c

m

Negative sign comes from comparator

For

s

M

s Y

m

· ) (

1

1

]

1

¸

+ τ

+ +

τ

−

·

1

1

]

1

¸

+ τ

+ τ

τ

−

· ′

1 ) 1 (

1

) (

2 2

s

C

s

B

s

A M K

s s

s M K

s P

f I

c

f

I

I

c

The

1 + τ s

C

f

term gives rise to an exponential.

To see the details of the response, we need to obtain B (= τ

I

- τ

f

) and A(=1)

by partial fraction expansion.

The response, shown for a negative change in Y

m

, would be

Note that as 0 →

τ

τ

I

f

, the two responses become the same.

d) If the measured level signal is quite noisy, then these changes might still

be large enough to cause the controller output to jump around even after

filtering.

One way to make the digital filter more effective is to filter the process

output at a higher sampling rate (e.g., 0.1 sec) while implementing the

controller algorithm at the slower rate (e.g., 1 sec).

A well-designed digital computer system will do this, thus eliminating the

need for analog (continuous) filtering.

time

y

-K

c

M

-K

c

M(1-τ

f

/τ

I

)

"Ideal" PI

Filtered PI

Slope = K

c

M/τ

I

8-9

8.8

a) From inspection of Eq. 8-26, the derivative kick = r

t

K

D

c

∆

∆

τ

b) Proportional kick = r K

c

∆

c) e

1

= e

2

= e

3

= …. = e

k-2

= e

k-1

= 0

e

k

= e

k+1

= e

k+2

= …= ∆r

p p

k

·

−1

1

]

1

¸

∆

∆

τ

+ ∆

τ

∆

+ ∆ + · r

t

r

t

r K p p

D

I

c k

1

]

1

¸

∆

τ

∆

+ + ∆ + ·

+

r

t

i r K p p

I

c i k

) 1 ( , i = 1, 2, …

c) To eliminate derivative kick, replace (e

k

– e

k-1

) in Eq. 8-26 by (y

k

-y

k-1

).

k

k+1 k+2

k+3

r

t

K

D

c

∆

∆

τ

r K

c

∆ r

t

K

I

c

∆

τ

∆

k

p

p

k-1

8-10

8.9

a) The digital velocity P algorithm is obtained by setting 1/τ

I

= τ

D

= 0 in Eq.

8-28 as

∆p

k

= K

c

(e

k

– e

k-1

)

= ( ) [ ]

1

) (

−

− − −

k sp k sp c

y y y y K

= [ ]

k k c

y y K −

−1

The digital velocity PD algorithm is obtained by setting 1/τ

I

= 0 in Eq. 8-

28 as

∆p

k

= K

c

[(e

k

– e

k-1

) +

t

D

∆

τ

(e

k

– 2e

k-1

+ e

k-2

)]

= K

c

[ (-y

k

+ y

k-1

) +

t

D

∆

τ

(-y

k

– 2y

k-1

+ y

k-2

) ]

In both cases, ∆p

k

does not depend on

sp

y .

b) For both these algorithms ∆p

k

= 0 if y

k-2

= y

k-1

= y

k

. Hence steady state is

reached with a value of y that is independent of the value of

sp

y . Use of

these algorithms is inadvisable if offset is a concern.

c) If the integral mode is present, then ∆p

k

contains the term K

c

) (

k sp

I

y y

t

−

τ

∆

.

Thus, at steady state when ∆p

k

= 0 and y

k-2

= y

k-1

= y

k

, y

k

=

sp

y and the

offset problem is eliminated.

8.10

a)

,

_

¸

¸

+ ατ

τ

+

τ

+ ·

′

1

1

1

) (

) (

s

s

s

K

s E

s P

D

D

I

c

( ) ( 1) 1

( 1)

I D D D I

c

I D

s s s s s

K

s s

τ ατ + + ατ + + τ τ

·

τ ατ +

1

]

1

¸

+ ατ τ

τ τ α + + ατ + τ +

·

) 1 (

) 1 ( ) ( 1

2

s s

s s

K

D I

D I D I

c

8-11

Cross- multiplying

( ) ) ( ) 1 ( ) ( 1 ) ( ) (

2 2

s E s s K s P s s

D I D I c I D I

τ τ α + + ατ + τ + · ′ τ + τ ατ

¸

¸

ατ + τ + ·

′

τ +

′

τ ατ

dt

t de

t e K

dt

t p d

dt

t p d

D I c I D I

) (

) ( ) (

) ( ) (

2

2

,

_

τ τ α + +

2

2

) (

) 1 (

dt

t e d

D I

b)

,

_

¸

¸

+ ατ

τ

,

_

¸

¸

τ

+ τ

·

′

1

1

) (

) (

s

s

s

s

K

s E

s P

D

D

I

I

c

Cross-multiplying

( ) ) ( ) 1 )( 1 ( ) ( ) 1 (

2

s E s s K s P s s

D I c D I

+ τ + τ · ′ + ατ τ

¸

¸

τ + τ + ·

′

τ +

′

τ ατ

dt

t de

t e K

dt

t p d

dt

t p d

D I c I D I

) (

) ( ) (

) ( ) (

2

2

,

_

τ τ +

2

2

) (

dt

t e d

D I

c) We need to choose parameters in order to simulate:

e.g., 2 ·

c

K , 3 · τ

I

, 5 . 0 · τ

D

, 1 . 0 · α , M = 1

By using Simulink-MATLAB

Step Response

Time

0 2 4 6 8 10

2

4

6

8

10

12

14

16

18

20

22

Parallel PID with a derivative filter

Series PID with a derivative filter

p'(t)

Figure S8.10. Step responses for both parallel and series PID controllers

with derivative filter.

8-12

8.11

a) ( ) 1

1

) (

) (

+ τ

,

_

¸

¸

τ

+ τ

·

′

s

s

s

K

s E

s P

D

I

I

c

( ) ) ( ) 1 )( 1 ( ) ( s E s s K s P s

D I c I

+ τ + τ · ′ τ

( ) ( )

( ) ( )

c

I D

I

K dp t de t

e t

dt dt

′

¸

· + τ + τ

τ

¸

,

_

τ τ +

2

2

) (

dt

t e d

D I

b) With the derivative mode active, an impulse response will occur at t = 0.

Afterwards, for a unit step change in e(t), the response will be a ramp

with slope = ( ) /

c I D I

K τ + τ τ and intercept = /

c I

K τ for 0 > t .

t

p '

slope =

Impulse at t=0

c

I

K

τ

( )

c I D

I

K τ + τ

τ