THE COPPERBELT UNIVERSITY

School of mines and minerals sciences

CE 560 / MT 580
Process control

Lecture 17
Transfer function development: Input-output process
models

C. Botha (Mr.)
Contacts: Chemical Engineering Department
Email: [email protected]
Alt. Email: [email protected]

May, 2023
 Transfer function

o The dynamic behaviour of a process can be determined by using two types of models:
a) Ordinary differential equation (ODE) and,
b) transfer function model; based on Laplace transform of input and output functions.
o Consider a process whose dynamic behaviour is described by an nth-order differential
equation (can be linear or non-linear).

d n y (t ) d n1 y (t ) dy (t )
an n
 an1 n 1
 ... a1  a0 y (t )  bf (t )...........................18.1
dt dt dt

o This process can be represented by a block diagram with a single input and output as
f (t ) y (t )
process
Input Output
 Transfer function development

o Block labelled process represents the dynamic model, Eq. 18.1

o Its corresponding block diagram with input and output in terms Laplace transforms is

Fig.16.1 (a) Process with input and output, (b) process represented by its transfer function

o Where 𝑓(𝑡) and 𝑦(𝑡) are the input and output variables of the process expressed in
deviation form.
o The block labelled 𝐺(𝑠) represents the transfer function model.
o Thus, transfer function relating the output to the input can be defined as
 Transfer function development

G ( s )  Transfer function
Laplace transform of output , in deviation form
Transfer function  ........(18.2)
Laplace transform of input , in deviation form
y (s)
G ( s)  .................................................(18.3)
f (s)

o Thus, 𝐺(𝑠) is characterised by dynamic relationship between dependent (output) and

independent (input) variables.
o Provides a much simpler input-output model than that given by the actual differential
equation (s).
o Allows complete description of the dynamic behaviour to be obtained when input is
known.
 Transfer function

o Therefore, for a particular change in input 𝒇 𝒕 , its transform 𝒇ത 𝒔 can be found and
from 18.3, the response is simply:
y ( s )  G ( s ) f ( s )...................................................(18.4)

o By taking the inverse transform of 18.4, the dynamic response 𝑦(𝑡) of a process is
obtained.
o Transfer function models are applicable to processes that exhibit linear dynamic
behaviour.
o That is, processes that can be modelled as linear ODE.
o And also non-linear processes provided the process can be linearised about steady state
operation.
 Transfer function development

Example: Transfer function for the stirred tank heater

o Process model for the stirred tank heater expressed in terms of deviation variables was
developed as follows:
dT  1 
 aT  Ti  KTst .......................................................(18.5)

dt 
o Where,
V UAi 1
  , K= and a   K
Fi V cp 
o Taking the Laplace transform of both sides of 16.5,

 dT  

 dt 
 aT  
1
 T   K
i Tst ..................................(18.6)
 Transfer function development: Stirred tank heater

o Since at steady state, 𝑇 ′ (0) = 0 then,

1
sT ( s )  T (0)  aT ( s )  Ti ( s )  KTst ( s )............................(16.7)

1 
sT ( s )  aT ( s )  Ti ( s )  KTst ( s ).................................(16.8)


1
T ( s )  s  a   Ti ( s )  KTst ( s )....................................(16.9)

1 K
T ( s )  Ti ( s )  Tst ( s ).............................(16.10)
s  a s  a
o Equation 16.10 describes the response of the process (stirred tank heater) to two inputs
 Transfer function development: Stirred tank heater

o These inputs are the inlet temperature and steam temperature.

o Defining the two transfer functions as:

T ( s ) T ( s )
G1 ( s )  and G2 ( s )  ..............................(16.11)
Ti( s ) Tst ( s )

o Thus, we can deduce from Eq. 16.10 that the two transfer functions are
T  (s) 1 T  ( s) K
G1 ( s )   and G2 ( s)   .............(16.13)
Ti ( s )  s  a  Tst ( s )  s  a 

o So that Eq. 16.10 can be written as

T ( s )  G1 ( s )Ti ( s )  G2 ( s )Tst ( s ).................................(16.12)
 Transfer function development: Stirred tank heater

o When the steam temperature remains unchanged, the transfer function for the stirred
tank heater is simply:
T  ( s) 1
G1 ( s )   ..........................................(16.14)
Ti ( s ) s  a
o The corresponding block diagram for the transfer function model (Eq. 16.10) is

o 𝐺1 (𝑠) represents the effect of inlet

temperature on temperature of
liquid in tank.
o 𝐺2 (𝑠) represents effect of steam
temperature on temperature of
liquid in the tank.
Fig.16.2 Block diagram for the stirred tank heater transfer function
 Transfer function development: Blending process

o Consider the isothermal blending process given in Fig. 16.3.

o The inlet stream 2 may not be

necessarily pure A. c A1 , w1 c A2 , w2

o The liquid level and hence tank hold-

up can change with time. h
cA
o The outlet flow-rate in this case is not
necessarily equal to inlet flow-rate. cA , w

o An input-out model is developed based Fig.16.3 Isothermal blending process

on a simplified case.
 Transfer function development: Blending process

o The fundamental quantity for the blending process is the mass ( both total mass and
mass of component A).
o Total mass balance
Rate of accumulation  Rate of inflow of  Rate of outflow of 
 =   ...........(16.15)
of total mass in the tank  mass into the tank  mass from the tank 

d V
= w1  w2  w........................................................(16.16)
dt

o Component A balance
Rate of accumulation  Rate of inflow of  Rate of outflow of 
     
of mass of A = mass of A   mass of A ..........(16.17)
in the tank  into the tank  from the tank 
     
 Transfer function development: Blending process
dc A V
= c A1 w1  c A2 w2  c A w..................................(16.18)
dt
o Assuming the following:
o Constant liquid density 𝜌 and volume 𝑉.
o Steady flow-rates i.e. constant 𝑤1, 𝑤2, and 𝑤.
o Equation 16.16 yields;
0  w1  w2  w.........................................(16.19)

dc A
o And 16.18 yields; V = c A1 w1  c A2 w2  c A w..................................(16.20)
dt
o When process is at steady state, 𝑐𝐴 = 𝑐𝑜𝑛𝑠𝑡 and
0  c A ,s w1  c A ,s w2  c A,s w....................................(16.21)
1 2
 Transfer function development: Blending process

o Subtracting 16.21 from 16.18 gives

d  c A  c A,s 
V =  c A  c A ,s  w1   c A  c A ,s  w2   c A  c A,s  w...(16.21)
dt 1 1 2 2

o Defining the following deviation variables;

cA1  c A1  c A1 ,s , cA2  c A2  c A2 ,s and cA  c A  c A,s .......(16.22)

o Then equation 16.21 yields;

dcA
V = cA w1  cA w2  cA w....................................(16.23)
dt 1 2

o Taking the Laplace transform of eq. 16.23

 dc
V  A  = w1 cA   w2 cA   w cA ......................(16.24)
 dt  1 2
 Transfer function development: Blending process

V  scA ( s )  cA (0)  = w1 cA ( s)   w2 cA ( s )   w cA ( s ) ...............(16.25)

1 2

o The tank is initially at steady state, hence; 𝑐 ′𝐴 (0) = 0, so that 16.25 becomes

V  scA ( s )   w cA ( s )  = w1 cA ( s )   w2 cA ( s )  ..........................(16.26)

1 2

cA ( s )  Vs  w = w1 cA1 ( s )   w2 cA2 ( s )  ......................................(16.27)

w1 w2
cA ( s )= cA1 ( s )  cA2 ( s )..........................................(16.28)
Vs  w Vs  w
w1 w2
cA ( s )= w c ( s )  w c ( s )..........................................(16.29)
V A1
V A2
s 1 s 1
w w
 Transfer function development: Blending process

o Defining the following parameters;   V , K  w1 and K  w 2

1 2
w w w
K1 K2
 cA ( s )= cA ( s )  cA ( s )...........................................(16.30)
 s 1 1
 s 1 2

o Defining the two transfer functions as

K1 K2
Gd ( s )= and G p ( s )  .....................................(16.31)
 s 1  s 1
o Then the transfer function for the blending process (eq. 16.30) can be written as
 cA ( s )= Gd ( s )cA1 ( s )  G p ( s )cA2 ( s )................................(16.32)

o The block diagram for the blending process is shown in figure 16.3.
 Transfer function development: Blending process
o As before, 𝐺1 (𝑠) 𝑜𝑟 𝐺d (𝑠) relates the
G1 ( s )
concentration of A in the tank to the inlet
concentration ഥ𝑐 ′𝐴1 (𝑠൯. cA1 ( s ) K1
 s 1
o As before, 𝐺2 (𝑠) 𝑜𝑟 𝐺p (𝑠) relates the +
cA ( s )
concentration of A in the tank to the inlet
concentration ഥ𝑐 ′𝐴2 (𝑠൯. G2 ( s ) +

cA2 ( s ) K2
o Fro the above two examples, its ease to deduce
 s 1
that the transfer function as a general form:
K Fig.16.4 Block diagram for the blending
G ( s) = ..........(16.32)
 s 1 process transfer function
o Where 𝜏 is the time constant and 𝐾 is the steady
state gain.
 Transfer function development: Blending process
Example 18.1

Consider the stirred tank blending process. The nominal steady state conditions are 𝑣1 =
10 𝐿Τ𝑚𝑖𝑛 , 𝑣2 = 20 𝐿Τ𝑚𝑖𝑛, 𝐶𝐴,1 = 1 𝑔Τ𝐿 and 𝐶𝐴,2 = 4 𝑔Τ𝐿 . The liquid volume and density
are constant: 𝑉=150 L and 𝜌 1200 kg/m3 respectively.

a) Calculate the nominal exit concentration, 𝐶𝐴ҧ

b) Derive an expression for the response 𝑥(𝑡) to a sudden unit step change in 𝑐𝐴,1 that occurs at
time 𝑡=0.

Assume that the process is initially operating at the nominal steady state.
 Transfer function development: General case for blending process
o When both concentrations and flow-rates change, LHS of equation 16.18 can be expanded as
d V dx
x +V = x1w1  x2 w2  xw..................................(16.33)
dt dt
o Where 𝑥1 , 𝑥2 , and 𝑥 are stream 1 and 2 concentrations and concentration of the product stream

o Combining Eq. 16.33 with a mass balance, eq. 16.16

dx
x( w1  w2 - w)+V = x1w1  x2 w2  xw...........................(16.34)
dt
o Simplifying
dx
V =  x1  x  w1   x2  x  w2 ................................(16.35)
dt
o If stream 2 consists of pure A, then 𝑥2 = 1 and
dx
V =  x1  x  w1  1  x  w2 ................................(16.36)
dt
 Transfer function development: General case blending process

o Eq. 16.36 is non-linear due to presence of the product terms such as 𝑥1 𝑤1 , 𝑥𝑤2 and so forth.

o Hence, RHS has a functional form 𝑓 𝑥, 𝑥1 , 𝑤1 , 𝑤2 .

o Procedure involves calculating nominal steady-state values 𝑤

ഥ and 𝑥ҧ by setting the derivates equal
to zero in total mass and component A balance.

o Next step is to linearize Eq. 16.36 about the nominal steady-state values calculated.

o Equation 16.36 expanded into Taylor series is

dx f f f f
V = x  x   x1  x1    w1  w1    w2  w2  ..........(16.37)
dt x s x1 s w1 s w2 s
 Transfer function development: General case blending process
o The partial derivates are as follows
f f f f
  w1  w2 ,  w1 ,  x1  x , and  1  x ..........(16.38)
x s x1 s w1 s w2 s

o Substituting equation 16.38 into 16.37 and defining deviation variables

dx 
V =  wx   w1 x1   x1  x  w1  1  x  w2 ...............................(16.39)
dt
o Taking the Laplace transform

Vsx  s  =  wx( s )  w1 x1( s )   x1  x  w1( s )  1  x  w2 ( s )...............(16.40)

o Rearranging and dividing by 𝑤
ഥ yields


x( s ) 
V  w1
s  1 =  x1( s ) 
 x1  x 
w1( s ) 
1 x 
w2 ( s )...............(16.41)
 w  w w w
 Transfer function development: General case blending process
o Defining the following


Vw1
, K1  , K 2 
 x1  x 
and K 3 
1 x 
...............(16.42)
w w w w
o Then 16.41 becomes

K1 K2 K3
x  s  = x1( s )  w1( s )  w2 ( s )...............(16.42)
 s 1  s 1  s 1

o Then, the individual transfer functions are as follows

x  s  K1 x  s  K2 x  s  K3
G1 ( s )  = , G2 ( s)   , and G3 ( s)   .....(16.42)
x1( s )  s  1 w1( s )  s  1 w2 ( s )  s  1
 Transfer function development: Blending process

Question 18.2

Consider the stirred tank blending process shown

below. The nominal steady state conditions are
c A1 , w1 c A2 , w2
𝑤1 = 600 kg/min , 𝑤2 = 20 kg/min , 𝑐𝐴1 =
0.050 and 𝑐𝐴2 = 1 pure A . The liquid volume
and density are constant: 𝑉 = 2 m3 and 𝜌 =
h
900 kg/m3 respectively. cA

a) Calculate the nominal exit concentration, 𝑐ҧA cA , w

b) Derive an expression for the response 𝑥 𝑡 to

a sudden change in both 𝑐𝐴1 and 𝑤1 from
𝑐𝐴1 = 0.050 to 0.075 and 𝑤1 =
600 to 650 respectively that occurs at time
𝑡 = 0. Assume that the process is initially at
the nominal steady state.
 Transfer function development: Blending process

Question 18.3

The stirred tank heating process is described by the input-output model (eq. 1) and
operates at steady state with an inlet temperature of 70 oF and a heat input of 1920
Btu/min. The liquid flow-rate is 200 Ib/min, the liquid has a constant density (𝝆 =
𝟔𝟐. 𝟒 𝑰𝒃/𝒇𝒕𝟑) and a specific heat (𝒄𝒑 = 𝟎. 𝟑𝟐 𝑩𝒕𝒖/𝑰𝒃 𝒐𝑭) and the liquid volume is
constant at 1.6 𝒇𝒕𝟑. Then the inlet temperature is changed to 90 oF and the heater input
changed to 1600 Btu/min. Calculate the output temperature response.

1 wc p 1
T ( s )  Q ( s )  Ti( s )..................(1)
m m
s 1 s 1
w w
 Additional questions

Question 18.4
q1
Suppose that the two liquid surge tanks are placed in series
so that the outflow from the first tank is the inflow to the
second tank. See figure 18.5. If the flow-rate from each tank
is proportional to the liquid (head) in that tank, find h1 R1
a) the transfer function relating changes in flow-rate from q1
the second tank 𝑄′ 2 (𝑠) to changes in flow-rate into the
first tank 𝑄 ′1 𝑠 . Assume that the two tanks have
different cross-sectional areas 𝐴1 and 𝐴2 and valve h2
resistances 𝑅1 and 𝑅2 .

b) The transfer function 𝐻 ′ 2 𝑠 / 𝑄 ′1 𝑠 . q2

c) Assuming that 𝐴1 = 𝐴2 and 𝑅1 = 𝑅2, determine the
response𝐻2 𝑡 to a unit step change in 𝑞1.
 Additional questions
F1, T1

Question 18.5

For the two stirred tank heaters connected in (𝒂) h1

A1
series as illustrated in the figures (a) and (b); F2, T2

a) Develop the state equation (model) Steam

h2
describing the dynamic behaviour of the
Q1
A2
system. F3, T3

b) Using the state model in (a), develop Steam

Q2

the transfer function for this process. F1

h1 h2
(𝒃) A1 A2
F2

Tank 1 Tank 2 F3
 Additional questions

Question 18.6

A stirred tank blending process with a constant liquid holdup of 2 m3 is used to blend two stream whose
densities are both approximately 900 kg/m3 the densities does not change during mixing.

a) The process has been operating for a long period of time with a flow-rate 𝑤1 = 500 𝑘𝑔/𝑚𝑖𝑛 and
𝑤2 = 200 𝑘𝑔/𝑚𝑖𝑛 and feed compositions (mass fractions) of 𝑥1 = 0.4 and 𝑥2 = 0.75. what is the
steady-state value of 𝑥?

b) Suppose that 𝑤1 changes suddenly from 500 to 400 kg/min and remains at the new value. Determine
an expression for 𝑥(t) and plot the results.

c) Repeat part (b) for the case where 𝑤2 changes (rather than 𝑤1 ) suddenly from 200 to 100 kg/min and
remains there.

d) Repeat part (c) for the case where 𝑥1 changes suddenly from 0.4 to 0.6.
 Additional questions

Question 18.7

Sketch the level response for a bathtub with cross-sectional area 8 ft3 as a function of time for the
following sequence of events, assume an initial level of 0.5 ft with the drain open. The inflow and outflow
are initially equal to 2 ft3/min.

a) The drain is suddenly closed and the inflow remains constant for 3 min (0 ≤ 𝑡 ≤ 3).

b) The drain is opened for 15 min; assume a time constant in a linear transfer function of 3 min, so a
steady state is essentially reached (3 ≤ 𝑡 ≤ 18).

c) The inflow rate is doubled for 6 min (18 ≤ 𝑡 ≤ 24)

d) The inflow rate is returned to its original value for 16 min 24 ≤ 𝑡 ≤ 40 .