Lecture 3

Process Dynamics:
Second and Higher Order Systems

Process Control L03-1

 Second order systems

• Output is described by the solution of second order

differential equations
• We find them in multi-capacity processes: two or more
capacities (first order systems) in series
• Inherently second order systems: fluid or mechanical
solid components of a process that possess inertia and
subjected to acceleration
• A processing system with its controller: controller
dynamics coupled with dynamics of the process unit

Process Control L03-2

 Second order systems: Examples

1. Inherently second order

• manometers
• pneumatic valves
• differential pressure transducers

2. Combinations of first order systems

• cascades of processes
• coupled systems

3. Systems constructed by humans

• process control systems

Process Control L03-3

 Second order systems

Time domain Laplace domain

f(t) y(t) F(s) Y(s)

Process GP(s)
Input Output Input Output

d2 y dy Kp
J2
+ 2.J + y = Kp · f(t) GP (s) = 2 2
dt2 dt J s + 2.Js + 1

Kp : steady state gain

J : process time
. : damping factor

Process Control L03-4

 Remarks

• As damping factor increases, second order effects (i.e. J2s2)

are far less important than first order effects (i.e. J2s2 << 2.Js)
the system behaves like first order

• As time passes by, second order effect diminish ( J2s2 6 0 )

the system behaves like first order

• Second order dynamics show differences when second order

effects (J2s2) are comparable or larger than first order effects
different behaviour for small .

Process Control L03-5

 Poles and damping factors
1
For a unit step change: f (t ) = 1; F ( s) =
s
Kp
y ( s) =
s(τ 2 s 2 + 2 ζτ s + 1)

In terms of poles: Kp / τ 2
y ( s) =
s( s − p1 )( s − p2 )

ζ ζ2 −1 ζ ζ2 −1
p1 = − + ; p2 = − −
τ τ τ τ
1) Damping factor > 1: two distinct real poles
2) Damping factor = 1: two equal poles
3) Damping factor < 1: two complex conjugate poles
L
Process Control L03-6
 Second order response

Step changes in the input

T
Output steady state Output

A
C
.1

.2 B

time time

Underdamped systems (. < 1)

Overshoot: A/B
Decay ratio: C/A
Period of oscillation: T

Process Control L03-7

 Characteristics of underdamped response

• The underdamped response is initially faster than the

critically damped or overdamped that are characterised
as sluggish
• Oscillates with progressively decreasing amplitude: a
distinct feature
• Oscillatory behaviour becomes more pronounced with
smaller values of damping factor

Process Control L03-8

 Jacketed reactor

FA,TA,CAO FB,TB,CBO

Coolant (Water)
FW,TW
F,T

What is the dynamic response of T with changes in TW?

Process Control L03-9

 Basic systems in jacketed reactor
FA, TA, CAO FB, TB, CBO

Tw Tjacket T

F,T

Reactor jacket
Chemical reactor
TW (induces changes to) Tjacket Tjacket (induces changes to) T

~ first order ~ first order

Process Control L03-10

 Changes of T with TW

(Reactor jacket) + (Chemical reactor)

KP,1 KP,2
Tjacket (s) = T (s) T (s) = T (s)
J1s + 1 W J2s + 1 jacket

Second order system

KP,2 KP,1
T (s) = · T (s)
(J2s + 1) (J1s + 1) W

reactor jacket

Process Control L03-11

 Remarks

• Chemical processes often involve combination of two or

more different processes
• The different systems may

mutually interact with support “one-way”

each other interactions
interacting non-interacting
systems systems

Process Control L03-12

 Can you identify the differences between the two
systems

Fin, 1

F1

h F1 = Fin, 2

h1 F2 h2

F2
h1(s) = G1 @ F1(s) h2(s) = G2 @ F2(s)

Process Control L03-13

 Non interacting systems
Fin, 1

K1
G1 = first order
(1) h F1 = Fin, 2
J1s + 1
G = G1·G2
(second order)
K2
(2) F2 G2 = first order
J2s + 1

Cascade of agitated tanks (1) and (2)

• (1) feeds (2) but
• (2) does not affect (1)
K1 K2
F2 (s) = · Fin, 1 (s)
J1s + 1 J2s + 1

Process Control L03-14

 Non interacting systems

K1 K2
G ( s) =
τ 1s + 1 τ 2 s + 1
K
G ( s) =
τ 2 s 2 + 2ξτs + 1
τ 2 = τ 1τ 2 ;2ξτ = (τ 1 + τ 2 ); K = K1 K2

The two poles of the transfer functions are real and distinct

1 1
p1 = − ; p2 = − Therefore, non-interacting capacities
τ1 τ2 always result in an overdamped or
critically damped second order system
and never in an underdamped system

L
Process Control L03-15
Dynamics of chemical processes: General case

K1 K2 Kn
GTOTAL = · ··············
J1 s + 1 J2 s + 1 Jn s + 1

basic basic basic

process 1 process 2 process n

How does this system behave with

respect to the basic processes?

Process Control L03-16

 Jacketed reactor
K1 .
K2
G=
J1 s + 1 J2 s + 1

Jacket Reactor vessel

Even if the jacket’s dynamics is extremely fast (i.e. J1 6 0):

K2
G = K1 ·
J2 s + 1
at best, the process is as fast as the slowest sub-process
(dynamics now determined by the reactor vessel)
in general the process is slower than both sub-processes
the combined dynamics yield an overdamped response

Process Control L03-17

 Interacting systems
F2
F1

h1 h2
• (1) feeds (2)

h1(s) = G1 @ F1(s) h2(s) = G2 @ F2(s) F • (2) resists outflow of (1)

K1 K2
G1 = G2 =
J1s + 1 J2s + 1
Interacting capacities
first order first order
(1) (2)

G = G1·G2
(second order)

always slower than non interacting

Process Control L03-18

 Interacting systems
dh1
A1 = F1 − F2
F1 dt
dh2
A2 = F2 − F
dt
1 F2 h2 h1 − h2 h2
F2 = ;F =
h1(s) = G1 @ F1(s) h2(s) = G2 @ F2(s) F R1 R2
dh1
A1 R1 + h1 − h2 = R1 F1
dt
dh2 R2 R2
A2 R2 − h1 + (1 + )h2 = 0
dt R1 R1
The two equations need to be solved simultaneously
The poles of the transfer function are real and distinct
Response of interacting capacities is always overdamped
Due to the interaction between the two capacities in both ways, interacting
capacities are more sluggish than the noninteracting
L
Process Control L03-19
 Higher order systems

Sub-process 1 noninteracting

Interacting

Sub-process 2 or interacting
noninteracting

Overdamped response
• general curve (no delays)
Sub-process 3

L
Process Control L03-20
 Delays on dynamics

f(t) f(t)

td
t t
Delay

Original signal... ... is equally available, but

only later

Consequently, the transfer ...is equally applicable,

function G(s)... but with some delay e-t s:
d

G(s) · e-t s d

Process Control L03-21

 Delays on combined processes
Output

Output
Sub-process 1

(delay)1
Input
Output

Overall
(delay)
Sub-process 2
Input
(delay)2 Input
Output
Overdamped response
with delay
Sub-process 3

(delay)3 Input

Process Control L03-22

 Process response: typical case

Output Output

K
td
– K
Js + 1 Y G=
Js + 1
· e-t d s

Input Input
td first order delay

Overdamped with delay First order with delay

Can be developed Simple calculations are Simple model

experimentally only required easily available

Process Control L03-23

 Development of a dynamic model

Step 1 : Identify input and output variables

Step 2 : Select an input variable and introduce a step change

Step 3 : Select an output variable and trace its response

(physical experiments, dynamic simulations)

Step 4 : Develop a first order model with delay to approximate

the response

Process Control L03-24

 Distillation Example*

XD Input : reflux R
R vapour boilup V
F
Output: XB, XD

F : feed

XB
V
Develop a dynamic model

* Wood and Berry, Chem. Eng. Sci, 1973, 28, p 1707

Process Control L03-25

 Input - output model

R XD
Distillation
V XB

• F is a fixed input
• To address the problem we need
- an experimental device, or
- a simulation model

Process Control L03-26

1. Introduce step changes to the reflux R (keep V constant)
Reflux

Time
2. Trace response of XD
XD

12.8
12.8 XD(s) – e-s · R(s)
16.7s + 1
16.7s + 1

1 Time

Process Control L03-27

 Similar experiments

R vs XDB
V V vs XBD
R V vs XBBD

6.6 e-7s -18.9 e-3s -19.4 e-3s

G=
10.9s + 1 21s + 1 14.4s + 1

Process Control L03-28

 Dynamic models : compact form

-3s
12.8 e-s -18.9
6.6 e-7s
XD 16.7s + 1 21 + 1
10.9s R
=
-7s
XB 6.6 e-3s
-18.9 -19.4 e-3s V
10.9
21s + 1 14.4s + 1

Outputs Dynamic models Inputs

Process Control L03-29

 Remarks

• Approximate model “hides” high order dynamics

• Overdamped response is the most common case, but

not the only case!

Process Control L03-30

 Linearisation of nonlinear models
The material and energy balance models that describe the behaviour of chemical
processes are nonlinear. Control strategies can become very complex and
challenging for non-linear systems. It is important, then, to be able to linearise
nonlinear models for control system design and analysis purposes. We use
Taylor series expansion for this purpose.
dx
= f ( x)
dt
∂f ( x ) 1 ∂ 2 f ( x)
f ( x ) ≈ f ( xs ) + ( x − xs ) + ( x − xs ) 2 + ...
∂x x s
2 ∂x 2
x
Using Taylor series expansion
s

∂f ( x )
f ( x ) ≈ f ( xs ) + ( x − xs ) Neglecting higher order terms
∂x x s

dx ∂f ( x )
= ( x − xs )
dt ∂x xs While using deviation variables, steady state value is zero
d ( x − xs ) ∂f ( x )
= ( x − xs )
dt ∂x xs
dx ' ∂f ( x )
= ax '; a = x’ is a deviation variable
dt ∂x xs

Process Control L03-31

 Taylor series expansion for linearisation
dx
= f ( x, u)
dt
∂f ( x , u) ∂f ( x , u)
f ( x , u ) ≈ f ( xs , us ) + ( x − xs ) + ( u − us )+...
∂x x ,u
s s
∂u x ,us s

∂f ( x , u) ∂f ( x , u)
f ( x , u ) ≈ f ( xs , us ) + ( x − xs ) + ( u − us ) Neglecting higher
∂x x ,u
s s
∂u x ,us s
order terms

dx ∂f ( x , u ) ∂f ( x , u)
= ( x − xs ) + ( u − us ) The s.s. value of f is zero
dt ∂x xs ,us ∂u xs ,us
d ( x − x s ) ∂f ( x , u) ∂f ( x , u)
= ( x − xs ) + ( u − us )
dt ∂x xs ,us ∂u xs ,us
dx ' ∂f ( x , u) ∂f ( x , u) In terms of deviation variables
= ax '+ bu'; a = ,b =
dt ∂x xs ,us ∂u xs ,us

Process Control L03-32

 Derive the linear models for a single, second order
reaction in a CSTR
dCA F
= f ( CA , F ) = ( CA0 − CA ) − kCA2
dt V
lt mol lt mol
k =1 , V = 1lt , CA0 = 2 , Fs = 1 , CAs = 1
mol.min lt min lt

Process Control L03-33

Results of linearisation for a single, second order
reaction in a CSTR

∂f (C A , F ) Fs
a= =− − 2 kC As
∂C A C As , Fs
V

∂f (C A , F ) C Ain − C As
b= = s

∂F C As , Fs V

a = −1 − 2 = −3 min −1
2 −1 mol
b= =1 2
1 lt
dC A '
= −3C A '+ F '
dt

Process Control L03-34