CHEMICAL PROCESS CONTROL

(20CGC047)

Semester 1 2021 (1b) Exam paper

This is a (1b) remote assessment examination, meaning you have a total of 2 hours plus an
additional 30 minutes to complete and submit this paper. The additional 30 minutes are for
downloading the paper and uploading your answers when you have finished. If you have
extra time or rest breaks as part of a Reasonable Adjustment, you will have further additional
time as indicated on your exam timetable.
You may handwrite and/or word process your answers, as you see fit.
You may use any calculator (not just those on the University’s approved list).
Attempt TWO questions: ONE from Section A and ONE from Section B
Each question carries 25 marks
A table of useful formulae and data is attached (Vade Mecum), pages 6 - 12
SECTION A – attempt ONE question from this section

Q1 The catalyst for a chemical reaction 𝐴 → 𝐵 is packed in a basket (Tank 2) through

which liquid circulates from Tank 1, as shown in Figure Q1. The feed to Tank 1 is a
dilute solution with a constant volumetric flow rate of 𝑞 2m h and a
concentration of 𝐶 𝑡 ; product leaves Tank 1 with a concentration of 𝐶 𝑡 . A
constant volumetric flowrate of 2𝑞 is fed into Tank 2, where the reaction proceeds at
a rate of 𝑟 20𝐶 ⁄ 1 2𝐶 kmol m h ; no reaction occurs in Tank 1. Both
tanks are well-mixed and have constant volumes, with 𝑉 16𝑉 4.0 m .

𝑉
𝑞 m h 𝑞 m h

𝐶 𝑡 kmol m Tank 1 𝐶 𝑡 kmol m

𝑞   2𝑞
𝐶 𝑡
𝑉 Tank 2
catalyst
Figure Q1 basket

Q1 continues on next page…

CGC022 / CGC047 1
Q1 continued…

(a) Noting any assumptions applied and using unsteady-state total mass balances,
express 𝑞 and 𝑞 in terms of 𝑞 .
[3 marks]

(b) Initially the system operates at steady state with 𝐶 0.500 kmol m . Show that
the initial steady state concentrations are 𝐶 0.250 kmol m and 𝐶
0.125 kmol m .
[5 marks]

(c) Using unsteady state material balances on 𝐴, show that the linearised system ODEs
are (* represents a deviation variable):

∗ ∗
0.5𝐶 ∗ 1.5𝐶 ∗ 𝐶∗ and 16𝐶 ∗ 28.8𝐶 ∗

[6 marks]

(d) Using the ODEs in part (c), show that the transfer function for this system is

#
# 𝐺 𝑠

[4 marks]

(e) Find the poles of the transfer function and comment on the system stability.
[2 marks]

(f) At 𝑡 0, there is a disturbance to the feed concentration, which can be modelled as

𝐶 𝑡 𝐶 0.5𝛿 𝑡 kmol m , where 𝛿 𝑡 is the Dirac delta function. What is the
maximum value of the concentration 𝐶 𝑡 , in the product stream?
[5 marks]

Q2 A second-order process has a transfer function given by

#
𝐺 𝑠 # (all time units are min)

(a) For a unit step change in the input variable, 𝑥 ∗ 𝑡 and using the information in section
2.2 of the Vade Mecum, calculate the following quantities and hence sketch the
output response:

Q2 continues on next page…

CGC022 / CGC047 2
Q2 continued…

(i) the new steady-state value of 𝑦 ∗ 𝑡 ;

(ii) the maximum value of 𝑦 ∗ 𝑡 ;

(iii) the decay ratio.

[6 marks]

(b) This process is to be controlled using a PI feedback system with a proportional gain
of 𝐾 and an integral reset time of 𝜏 . The measurement sensor and valve have
transfer functions

𝐺 𝑠 and 𝐺 𝑠 1

Sketch a block diagram of the feedback control loop, labelling all streams and
showing all transfer functions.
[3 marks]

(c) Find the open loop transfer function and show that the characteristic equation for the
feedback control system may be written as

9𝑠 27𝑠 22𝑠 8 1 𝐾 𝑠 8𝐾 /𝜏 0
[4 marks]

(d) Apply the Routh test (section 7 of the Vade Mecum) to the characteristic equation for
P control only (hint: allow 𝜏 → ∞) and hence find the critical gain and period of
oscillation.
[5 marks]

(e) Hence suggest settings for a PI controller using the Ziegler-Nichols rules.
[2 marks]

(f) Use the Routh array to find the minimum integral reset time that could be used with
the proportional gain calculated in part (e).
[5 marks]

END OF SECTION A

CGC022 / CGC047 3
 SECTION B – attempt ONE question from this section

Q3 Figure Q3 shows a blending system in which the output concentration is controlled by

manipulating the feed concentration 𝐶 directly. The output concentration 𝐶 is
measured at the end of a long pipe of length 𝐿 30 m, and cross-sectional area 𝐴
0.05 m . The flow rate is 𝑞 0.3 𝑚 min , and the volume in the blending tank is
constant and equal to 𝑉 2.1 m . The dynamics of the sensor and control valve are
negligible.

Pure A
AC
𝑞, 𝐶𝐴𝑓
Solvent
Figure Q3

𝑉, 𝐶𝐴 𝐿 𝐶𝐴𝑜

(a) Draw the block diagram of the system. Label every signal and identify every block.
[2 marks]

(b) Consider that the pipe can be modelled by a pure delay system. Show that the
process model (blending vessel with pipe) has the following transfer function:

#
𝐺 𝑠 #

[5 marks]

(c) Using a first-order Padé approximation of the time delay, design a PI controller using
the Ziegler-Nichols method.
[9 marks]

(d) Using the Internal Model Control (IMC) design procedure with a first-order filter with a
time constant 𝜆 2.5, design a controller for the process. Derive the expression for
the equivalent feedback controller and give the parameters of the PID controller
which is equivalent with the IMC structure obtained.
[9 marks]

CGC022 / CGC047 4
Q4 The transfer function model of a certain chemical process is given as:

𝐺 𝑠

(a) By considering the poles and zeros of the overall transfer function, indicate whether
the system is stable, overdamped/underdamped and/or shows inverse response.
Sketch the step response of the system.
[3 marks]

(b) Sketch the Bode diagram of the process. Identify on the sketch all important
distinguishing characteristics (consider high and low frequency asymptotes and
resonance). Derive the amplitude ratio and phase angle for the process transfer
function.
[4 marks]

(c) Under pure proportional control what is the range of values that the controller gain 𝐾
can take for the closed-loop system to be stable? Assume there are no other
elements in the feedback loop, just the process and the feedback controller.
[4 marks]

(d) Consider the case when a minor loop around the pure proportional controller is used
as shown in Figure Q4.

+ +
𝐾 𝐺
_ _

𝐺 Figure Q4

The transfer function in the minor loop is chosen to be:

𝐺 𝑠

Find the range of values which 𝐾 can take for the overall system to be stable.
[7 marks]

(e) If the pure proportional controller and the transfer function in the minor loop around it
(𝐺 ) are combined into one apparent controller block, 𝐺 , show that in the limit as
𝐾 → ∞ , this apparent controller tends to a PID controller. What are the values for
the gain, integral time and derivative time for this controller?
[7 marks]

END OF PAPER
Chris D. Rielly, Brahim Benyahia
CGC022 / CGC047 5
Department of Chemical Engineering

A CONTROL ENGINEER'S
VADE MECUM1

 
1.   LAPLACE TRANSFORMATIONS ................................................................... 7 
1.1  Properties & simple functions ............................................................................ 7 
1.2  Convolution ....................................................................................................... 7 
1.3  Initial and final-value theorems .......................................................................... 7 
2.  SECOND-ORDER SYSTEM RESPONSES .................................................... 8 
2.1  Second order transfer functions ........................................................................ 8 
2.2  Characteristics of the underdamped response to a unit step input ................... 8 
3.   BLOCK DIAGRAM TRANSFORMATIONS ...................................................... 9 
4.  FREQUENCY RESPONSE ........................................................................... 10 
4.1  Ziegler-Nichols controller settings ................................................................... 10 
4.2  Amplitude ratio and phase lag of various systems .......................................... 10 
5.  PADÉ APPROXIMATION .............................................................................. 10 
6.  LINEARISATION ........................................................................................... 11 
6.1  Single variable functions ................................................................................. 11 
6.2  Multi-variable functions .................................................................................... 11 
7.  ROUTH TEST FOR STABILITY .................................................................... 11 
8.  CONVENTIONAL FORM OF IMC CONTROLLER ........................................ 12 

1
A body of knowledge which can be brought to bear quickly (without the need of redemonstration)
on the problem in hand.

CGC022 / CGC047 6
1. Laplace transformations
1.1 Properties & simple functions


Definition Lf (t )  f (s )  f (t )e  st dt
#

0
f (t ) f # (s ) f (t ) f # (s )
1 s
1 cos ( t )
s s  2
2

1 1
t t e  t
s2 (s   )2
Simple Functions

n! 
tn e  t sin(t )
s n 1 (s   )2   2
1 s 
e t e  t cos(t )
s  (s   )2   2
  k  exp(k s )
sin(  t )
2 2
erfc 
s  2 t  s

H (t )
1
H (t   ) e s
s s
 (t ) 1  (t   ) e s
Shift in t Shift in s
Properties

H(t   )f (t   ) e s #
f (s ) e  t f (t ) f # (s   )
Scaling f # (s /  ) Linearity
f ( t )  af t   bg t  af # s   bg # s 
t
df ( t ) f # s 
#
sf (s )  f (0)  f ( ) d 
Derivatives and

dt s
0
Integrals


df # (s ) f (t )
f
#
tf (t )  ( )d 
ds t
s
n
d f (t )
s n f # (s )  s n 1f (0)  s n  2f (0)  ...  f ( n 1) (0)
n
dt

1.2 Convolution

 t 

L  f1( )f2 (t   ) d   f1# (s )f2# (s )
 0 

1.3 Initial and final-value theorems

lim f (t )  lim sf # (s ) lim f (t )  lim sf # (s )

t 0 s  t  s 0

CGC022 / CGC047 7
2. Second-order system responses
2.1 Second order transfer functions

The general second-order system a0  0

d 2y dy
2  2  y  K p x t 
2 dt
dt
where  = time constant, where 2 = natural period of oscillation = Tn,
 = damping factor and
Kp = steady-state gain.

has a transfer function

y# Kp
Gs   
# 2 2
x  s  2s  1

2.2 Characteristics of the underdamped response to a unit step input

tr t95

t / 
 
A  
Overshoot  exp  
B  1  2 
 

  2
C  2    A 
Decay ratio  exp 
A  1  2  B
 
2 2
Period T  
 1  2

tr is the rises time and t95 is the response time (defined here as the time for the
oscillation to decay within 5% of the new steady-state value).

When  = 0, the output oscillates continuously, with a natural period Tn  2 .

CGC022 / CGC047 8
3. Block diagram transformations

In the following table, G, G1 andG2 are transfer functions

Equivalent block
Transformation Equation Block diagram
diagram

Combining
x# y# x# y#
1 blocks in y #  G1G2 x # G1 G2 G1 G2
series

x# + y#
G1
Combining
 x# y#
2 blocks in y #  G1x #  G2 x # G1  G2
parallel G2

x# + y#
G1
Eliminating x# y#
3 a feedback y #

 G1 x  G2 y# #
  G1
1  G1G2
loop G2

w# + + z# w# + + z#
Rearranging
4 summing z#  w #  x #  y #    
x# y#
junctions
y# x#

Moving a x# + z#
summing x# + z# G
G
5 junction z #  Gx #  y # 

ahead of a 1 y#
y#
G
block

Moving a x# + z#
summing x# + z# G
 
G
6 junction z#  G x#  y # 

beyond a y#
y# G
block

Moving a x# y# x# y#
take-off G G
7 ahead of a y #  Gx #
y# y#
block G

Moving a x# y# x# y#
take-off G G
8 beyond a y #  Gx #
x# x# 1
block G

CGC022 / CGC047 9
4. Frequency response
4.1 Ziegler-Nichols controller settings
Controller Transfer Controller ntegral Derivative Time
Type Function Gain Time
1
P Kc K c crit – –
2
 1  9 5
PI K c 1   K c crit Pcrit –
 Is  20 6
 1  3 1 1
PID K c 1    Ds  K c crit Pcrit Pcrit
 Is  5 2 8

Legend: K c crit = critical controller gain,

Pcrit = period of oscillations at critical conditions

4.2 Amplitude ratio and phase lag of various systems

Transfer function Amplitude ratio Phase Angle
System Gs  AR  G i    argGi 
Time
Gd s   e  t d s ARd  1 1  td
delay

Pure lead Gs 1 s AR  1   2 2   tan 1 

K K
First G1s   AR1  1  tan 1   
order s  1 1  2 2

Second K K   2 
G2 (s) 
2 2
AR 2 
2 2 2 2
2  tan1 
order  s  2s  1 (1    )  ( 2 )  1   2 2 

PI  1  K c 1  I2 2
GI s   K c 1   ARI  I  tan 1 I     / 2
Controller  Is   I

PD GD s   Kc 1   Ds  2 2
ARD  K c 1   D D  tan 1 D  
Controller

5. Padé approximation
1  0.5 s
First-order Padé approximation e  s 
1  0.5 s

CGC022 / CGC047 10
6. Linearisation
6.1 Single variable functions
The Taylor series expansion about a steady-state condition xs is given by

f x   f x s   x  x s 
df

x  x s 2 d 2 f  ...
dx s 2! dx 2 s

6.2 Multi-variable functions

For a two variable non-linear function the Taylor series expansion about a steady-
state condition x1s , x2s  is given by

f f
f x1, x 2   f x1s , x 2s   x1  x1s   x 2  x 2s  
x1 s x 2 s

x1  x1s   2f  x1  x1s x 2  x 2s 

 2f

x 2  x 2s   2 f  ....
2! x12 s x1 x 2 2! x 22 s
s

7. Routh test for stability

The characteristic equation is written as an nth order polynomial in s (a0 > 0)
a0 s n  a1s n 1  a2s n  2    an  0

Test One: If any of the coefficients a0, a1, ..., an is negative, then there is at least
one positive real root to the characteristic equation and the system is
unstable.

If all of a0, a1, ..., an > 0, then proceed to Test Two.

Test Two: The next stage is to form the Routh array

Row 1 a0 a2 a4 a6 
2 a1 a3 a5 a7 
3 A1 A2 A3 A4 
4 B1 B2 B3 B4 
5 C1 C2 C3 
    
n  1 Z1 Z2 Z3 

a a a a a a  a0a5 a a  a0a7
where A1  1 2 0 3 , A2  1 4 A3  1 6
a1 a1 a1
A a a A A a  a1A3 A a  a1A4
B1  1 3 1 2 , B2  1 5 B3  1 7
A1 A1 A1
B A  A1B2 B A  A1B3 B A  A1B4
C1  1 2 , C2  1 3 C3  1 4
B1 B1 B1

CGC022 / CGC047 11
 Examine the first column of numbers: a0, a1, A1, B1, , Z1
 If any of these numbers is negative, then there is at least one root of the
characteristic equation which has a positive real part. The system is unstable.
 The number of sign changes in the first column is equal to the number of roots
with positive real parts.

At the critical gain, the n  1th row has coefficients which are and , in which case
the purely imaginary roots are given by

 s2    0

8. Conventional form of IMC controller

GI (s )
Gc (s )  
1  GI (s )Gm (s )

GI – transfer function of IMC controller

Gm – transfer function of the process model






CGC022 / CGC047 12