10/10/2021

Ali M. Sahlodin
Department of Chemical Engineering
AmirKabir University of Technology

Transfer functions

Transfer function properties and operations

Linearization of nonlinear models

1
 10/10/2021

Response of simple dynamic models to

standard inputs
 We use transfer functions

U(s) G1(s) Y(s)

Step

Ramp

Rectangle pulse

Impulse

Sinusoidal

Random

2
 10/10/2021

Periodic changes in process

usin variables

0 t0
U sin  t   
 A sin t  t0

A = amplitude
 = angular frequency

Examples?
•Daily changes in ambient temperature
•Pressure-swing adsorption (PSA)
5

Many types of inputs can be constructed from combination of

step, ramp, pulse, etc.

Process Control & Dynamics, Seborg et al.

3
 10/10/2021

 See the book example (also given in the class).

Triangle pulse input

Y ( s)
 G(s)
U (s) Y ( s )  G ( s)
U (s)  1

Impulse response is the process model itself!

4
 10/10/2021

First-order transfer function (standard form)

Y s K K  steady-state gain


U s τs  1 τ  time constant

Response to a step change of magnitude M

M
U  t   M for t  0  U s 
s
KM
Y s 
s  τs  1
9

KM
Y ( s) 
(τs  1) s


y  t   KM 1  e t /τ  y  KM
y
1.0
t y___

0 0
y
0.5 τ 0.632
y 2τ 0.865
3τ 0.950
0 4τ 0.982
0 1 2 3 4 5 0.993
5τ
t
τ
Note: Large τ means a slow response.
10

5
 10/10/2021

After a transient, the response will be the same

as the input shifted by the time constant.

Transient response

Process Control & Dynamics, Seborg et al.

11

Sine input U (t )  A sin(t )

Inverse Laplace

Or
A more
useful form

12

6
 10/10/2021

Ensures a nearly constant flow to downstream

units despite sudden changes/oscillations in the
upstream flow

The larger the tank, the larger the time
constant
 Better damping of the oscillations

Flow out

Flow in

13

qi

q
Pump (instead of valve) q  f h

dh 1
A  qi  q  0  qi  q H  s  Qi  s   Q  s  
dt As 
H  s  1
Q  s   0 
Qi  s  As

What assumptions did we make?

14

7
 10/10/2021

Transfer function
Y s K
  K  constant 
U s s

Step change of magnitude M

KM
Y s  y  t   KMt
s2

Any steady state?

15

Standard form Y s K


U s 2 2
τ s  2ζτs  1

K steady-state gain
τ time constant [time]
ζ damping coefficient (dimensionless)

1

Another form (natural frequency n  )


Y s K n2

U s s 2  2ζn s  n2
16

8
 10/10/2021

Two first-order processes in series

G ( s )  G1 ( s )G 2 ( s )

Vehicle
suspension system (inherently
second-order systems)

Adapted from wikihow

https://www.wikihow.com/Inspect-Your-Suspension-System

17

18

9
 10/10/2021

Overdamped

Critically damped

Underdamped

19

More interesting than first-order systems!

Behavior depends on value of zeta.

Characteristic polynomial τ 2 s 2  2ζτs  1

Damping Type of Response Roots of Charact.
Coefficient Polynomial
ζ 1 Overdamped Real and distinct

ζ 1 Critically damped Real and repeated

0  ζ 1 Underdamped Complex conjugates

Where do these conclusions come from?
20

10
 10/10/2021

Process Control & Dynamics, Seborg et al.

•Damping factor less than 1 leads to oscillations and overshoot

•Fastest response without overshoot: critically damped
•The larger the damping coefficient, the slower the response

21

Rise time (tr)

Time to first peak (tp)

Overshoot (OS=a/b)

Decay ratio (DR=c/a)

Settling time (ts)

Period of oscillation (P)

Process Control & Dynamics, Seborg et al.

Can you derive analytical expressions for these parameters?

How to identify the system using this response?

22

11
 10/10/2021

Underdamped process

Process Control & Dynamics, Seborg et al.

Can be used to obtain estimates of damping coeff.

23

Output oscillates with larger amplitude than input (undesired!)

Dynamics, Seborg et al.
Process Control &

Response after
decaying
exponential terms
(t  ∞)

Amplitude ratio depends on both damping coeff. and frequency

Underdamped systems with zeta<0.707: AR can be greater than 1
depending on frequency

Systems with zeta>0.707: always AR <1 (e.g., two tanks in series)
24

12