Transfer Function Models

Convenient representation of a linear, dynamic model.

A transfer function (TF) relates one input to one output:

x t y t
system
X s Y s

The following terminology is used:

x y
input output
forcing function response
cause effect

Definition of the transfer function:

Let G(s) denote the transfer function between an input, x, and an
output, y. Then, by definition
Y ( s)
G( s)
X ( s)
where:
Y ( s ) L y (t )
X ( s ) Lx (t )

Development of Transfer Functions

Example: Stirred Tank Heating System
 Figure 2.3 Stirred-tank heating process with constant holdup, V.

For the dynamic model, assuming constant liquid holdup and

flow rates:
dT
V C wC Ti T Q (1)
dt
Suppose the process is initially at steady state:

T 0 T , Ti 0 Ti , Q 0 Q 2
where T steady-state value of T, etc. For steady-state conditions:

0 wC Ti T Q (3)

Subtract (3) from (1):

wC Ti Ti T T Q Q
dT
V C (4)
dt
But,
dT d T T
because T is a constant (5)
dt dt

Thus we can substitute into (4-2) to get,

dT
V C wC Ti T Q (6)
dt
where we have introduced the following deviation variables,
also called perturbation variables:

T T T , Ti Ti Ti , Q Q Q ( 7)

Take L of (6):

V C sT s T t 0 wC Ti s T s Q s (8)

Evaluate T t 0 .

By definition, T T T . Thus at time t = 0,

T 0 T 0 T (9)

But since we assumed that the process was initially at steady

state, i.e., T 0 T , it follows from (9) that T 0 0.
Note: The advantage of using deviation variables is that the
initial condition term becomes zero. This simplifies the later
analysis.

Rearrange (8) to solve for T s :

K 1
T s Q s Ti s (10)
s 1 s 1
where two new symbols are defined:

1 V
K and (11)
wC w
Transfer Function Between Q and T
Suppose Ti is constant at the steady-state value. Then,
Ti t Ti Ti t 0 Ti s 0.Then we can substitute into (10)
and rearrange to get the desired TF:

T s K
(12)
Q s s 1

Transfer Function Between T and Ti :

Suppose that Q is constant at its steady-state value:

Q t Q Q t 0 Q s 0

Thus, rearranging

T s 1
(13)
Ti s s 1

Comments:
1. The TFs in (12) and (13) show the individual effects of Q and Ti
on T. What about simultaneous changes in both Q and Ti ?
 T (s)=G1(s)Q(s) G2(s)Ti(s)

G1 and G2 are transfer functions and independent of the

inputs, Q and Ti.
Note G1 (process) has gain K and time constant
Chapter 4

G2 (disturbance) has gain=1 and time constant

gain = G(s=0). Both are first order processes.
If there is no change in inlet temperature (Ti= 0),
then Ti(s) = 0.
System can be forced by a change in either Ti or Q
(see Example 4.3).

Answer: See (10). The same TFs are valid for simultaneous
changes.

Note that (10) shows that the effects of changes in both Q

and Ti are additive. This always occurs for linear, dynamic
models (like TFs) because the Principle of Superposition is
valid.
2. The TF model enables us to determine the output response to
any change in an input.
3. Use deviation variables to eliminate initial conditions for TF
models.
 Example: Stirred Tank Heater
K 0.05 2.0
0.05 No change in Ti
T Q
2s 1
Step change in Q(t): 1500 cal/sec to 2000 cal/sec
Chapter 4

500
Q
s
0.05 500 25
T
2s 1 s s (2 s 1)
What is T(t)? From line 13, Table 3.1
25
T (t ) 25[1 e t / ]
T (s)
s ( s 1)
T (t ) 25[1 e t / 2 ]
11

Properties of Transfer Function Models

1. Steady-State Gain
The steady-state of a TF can be used to calculate the steady-
state change in an output due to a steady-state change in the
input. For example, suppose we know two steady states for an
input, u, and an output, y. Then we can calculate the steady-
state gain, K, from:

y2 y1
K (4-41)
u2 u1

For a linear system, K is a constant. But for a nonlinear

system, K will depend on the operating condition u , y .
Calculation of K from the TF Model:
If a TF model has a steady-state gain, then:

K lim G s (14)
s 0

This important result is a consequence of the Final Value

Theorem

Note: Some TF models do not have a steady-state gain (e.g.,

integrating process in Ch. 5)

2. Order of a TF Model
Consider a general n-th order, linear ODE:
dny dy n1 dy d mu
an an1 a1 a0 y bm m
dt n dt n1 dt dt
d m1u du
bm1 b1 b0u (4-42)
dt m1 dt
Take L, assuming the initial conditions are all zero. Rearranging
gives the TF:
m

Y s i 0
bi si
G s (4-43)
U s n
ai si
i 0

n
Characteristic equation
a s
i 0
i
i
0 ( )
Definition:
The order of the TF is defined to be the order of the denominator
polynomial.
Note: The order of the TF is equal to the order of the ODE.

Physical Realizability:
For any physical system, n m in (4-38). Otherwise, the system
response to a step input will be an impulse. This cant happen.
Example:

du
a0 y b1 b0u and step change in u (4-45)
dt

3. Additive Property
Suppose that an output is influenced by two inputs and that
the transfer functions are known:
Y s Y s
G1 s and G2 s
U1 s U2 s
Then the response to changes in both U1 and U 2 can be written
as:

Y s G1 s U1 s G2 s U 2 s
The graphical representation (or block diagram) is:

U1(s) G1(s)
Y (s )

U2(s) G2(s)
 4. Multiplicative Property
Suppose that
Y s U2 s
G2 s and G3 s
U2 s U3 s
Then,
Y s G2 s U 2 s and U 2 s G3 s U 3 s
Substitute,
Y s G2 s G3 s U 3 s
Or,

Y s
G2 s G3 s U3 s G2 s G3 s Y s
U3 s

5. Poles and Zeros ( )

Suppose that

Y ( s) K ( s z1 )( s z 2 ) ( s z m )
G(s)
U ( s) ( s p1 )(s p 2 ) ( s p n )
Chapter 4

p1, p2, , pn poles (real or complex)

z1, z2, , zm zeros (real or complex)

p1, p2, , pn

U(s) = M s (Bounded input function)

A0 A1 An
Y ( s)
s s p1 s pn

y (t ) A0 A1e p1t A2 e p2t An e pnt

Ai

18
 (i)pi

t , y(t) grow exponentially without bound or diverge

(ii)
Chapter 4

t , y(t) A0 converge

Y ( s) 1
e.g. ; U ( s) 1
U ( s ) s p1

y (t ) e p1 t , p1
p1=0 p1>0
p1<0
y(t)

The response is monotonic (nonoscillatory)

19

Y ( s) 1
e.g.
U ( s ) ( s p1 )( s p2 )
; U ( s) 1 ; p1, 2 j

1
y (t ) e t sin t

Chapter 4

< 0 = 0 >0
y(t)

* The response is oscillatory

20
Linearization of Nonlinear Models
So far, we have emphasized linear models which can be
transformed into TF models.
But most physical processes and physical models are nonlinear.
- But over a small range of operating conditions, the behavior
may be approximately linear.
- Conclude: Linear approximations can be useful, especially
for purpose of analysis or control.
Approximate linear models can be obtained analytically by a
method called linearization. It is based on a Taylor Series
Expansion of a nonlinear function about a specified operating
point.

Consider a nonlinear, dynamic model relating two process

variables, u and y:
dy
f y, u (4-70)
dt
Perform a Taylor Series Expansion about u u and y y and
truncate after the first order terms,
f f
f (u , y ) f (u , y ) u y (4 71)
u y ,u y y ,u

where u u u and y y y . Note that the partial derivative

terms are actually constants because they have been evaluated at
the nominal operating point, u , y .
Substitute (4-71) into (4-70) gives:
dy f f
f (u , y ) u y
dt u y ,u y y ,u
The steady-state version of (4-70) is:

0 f u , y

Substitute into (7) and recall that dy dy

,
dt dt

dy f f
u y (4 72)
dt u s y s Linearized
model
Example: Liquid Storage System
dh
qi Mass balance: A qi q (1)
dt
Valve relation: q Cv h (2)

h A = area, Cv = constant
q

Combine (1) and (2),

dh
A qi Cv h (3)
dt

Linearize term,

1
h h (h h ) ( 4)
2 h
Or
1
Cv h Cv h (h h ) (5)
R

where: 2 h
R
Cv
h h h
Substitute linearized expression (5) into (3):

dh 1
A qi Cv h h ( 6)
dt R
The steady-state version of (3) is:

0 qi Cv h (7)

Subtract (7) from (6) and let qi qi qi , noting that dh dh

gives the linearized model: dt dt

dh 1
A qi h (8)
dt R

Summary:
In order to linearize a nonlinear, dynamic model:

1. Perform a Taylor Series Expansion of each nonlinear term

and truncate after the first-order terms.
2. Subtract the steady-state version of the equation.
3. Introduce deviation variables.
 State-Space Models

Dynamic models derived from physical principles typically

consist of one or more ordinary differential equations (ODEs).
In this section, we consider a general class of ODE models
referred to as state-space models.

Consider standard form for a linear state-space model,

x = Ax + Bu + Ed (1)

y = Cx (2)

where:
x = the state vector
u = the control vector of manipulated variables (also called
control variables)
d = the disturbance vector
y = the output vector of measured variables. (We use
boldface symbols to denote vector and matrices, and
plain text to represent scalars.)
The elements of x are referred to as state variables.
The elements of y are typically a subset of x, namely, the state
variables that are measured. In general, x, u, d, and y are
functions of time.
The time derivative of x is denoted by x dx / dt .
Matrices A, B, C, and E are constant matrices.
Example 4.10
Show that the linearized CSTR model of Chap. 2 can
be written in the state-space form of Eqs. (1) and (2).
Derive state-space models for two cases:
(a) Both cA and T are measured.
(b) Only T is measured.

Solution
The linearized CSTR model can be written in vector-matrix form:

dcA a a12 cA 0
dt 11 T
s
(3)
dT a a22 T b2
dt 21

Let x1 c A and x 2 T , and denote their time derivatives by x1

and x2 . Suppose that the steam temperature Ts can be manipulated.
For this situation, there is a scalar control variable, u Ts , and
no modeled disturbance. Substituting these definitions into (3)
gives,

x1 a11 a12 x1 0
x a x b u (4)
2 21

a22 2
2
A B

which is in the form of Eq. (1) with x = col [x1, x2]. (The symbol
col denotes a column vector.)
a) If both T and cA are measured, then y = x, and C = I in
Eq. (2), where I denotes the 2x2 identity matrix. A and B are
defined in (3).

b) When only T is measured, output vector y is a scalar,

y T and C is a row vector, C = [0,1].

Note that the state-space model for Example 3.10 has d = 0

because disturbance variables were not included in (4). By
contrast, suppose that the feed composition and feed temperature
are considered to be disturbance variables in the original
nonlinear CSTR. Then the linearized model would include two
additional deviation variables, cAi and Ti .