Office Use Only

Monash University

Semester Two Examination Period 2012

Faculty Of Engineering

EXAM CODES: CHE3162

TITLE OF PAPER: PROCESS CONTROL

EXAM DURATION: 3 hours writing time

READING TIME: 10 minutes

THIS PAPER IS FOR STUDENTS STUDYING AT:( tick where applicable)

 Berwick  Clayton  Malaysia  Off Campus Learning  Open Learning
 Caulfield  Gippsland  Peninsula  Enhancement Studies  Sth Africa
 Pharmacy  Other (specify)

During an exam, you must not have in your possession, a book, notes, paper, electronic device/s,
calculator, pencil case, mobile phone or other material/item which has not been authorised for the
exam or specifically permitted as noted below. Any material or item on your desk, chair or person
will be deemed to be in your possession. You are reminded that possession of unauthorised materials
in an exam is a discipline offence under Monash Statute 4.1.

No examination papers are to be removed from the room.

AUTHORISED MATERIALS

CALCULATORS  YES  NO
Only calculators with ‘approved for use’ Faculty of Engineering or Faculty of Science stickers are permitted

OPEN BOOK  YES  NO

SPECIFICALLY PERMITTED ITEMS  YES  NO

if yes, items permitted are:

Candidates must complete this section if required to write answers within this paper

STUDENT ID __ __ __ __ __ __ __ __ DESK NUMBER __ __ __ __

Page 1 of 23
 EXAM INSTRUCTIONS

ATTEMPT ALL FIVE (5) QUESTIONS

EQUATION SHEETS are provided

Write all answers in the exam book provided.

TOTAL MARKS for this exam equals 70.

Page 2 of 23
QUESTION 1 (15 marks):

You are interested in modelling a 10 m3 surge tank which holds the products exiting a plug flow
reactor, as shown below. One of the purposes of the tank is to heat the reactor products to the final
temperature T prior to the next stage of the process. The sides of the surge tank are surrounded by a
steam jacket which uses steam at temperature Ts. However, due to poor design, heat is lost from the
tank via heat transfer from the surface of the product liquid to the air above the tank. The inlet
temperature TIN also varies. The feed to the tank flows at 0.2 m3/s, and you can assume that the tank
is always full. You can assume that the following quantities are constant:

UA (steam jacket) = 250 kJ/K/s, UA (liquid-air interface) = 20 kJ/K/s

ρ = 1000 kg/m3, Cp = 4.2 kJ/kg/K.
Feed stream
air
Steam
Exit stream

Condensed
steam
(a) [6 marks] Using an energy balance, prove that this system can be described by three
transfer functions T’/T’IN, T’/T’A, and T’/T’S

T ' 0.76 T ' 0.018 T ' 0.23

= = =
TIN ' 38s + 1 TA ' 38s + 1 Ts ' 38s + 1

where T is the outlet temperature, TS is the steam temperature, TA is the air temperature, and
TIN is the inlet temperature.

(b) [3 marks] Draw a block diagram for T’ as a function of T’A, T’IN and T’S.

(c) [2 marks] Determine the steady state temperature in the surge tank if the air temperature
is 20oC, the steam temperature is 110oC and the inlet temperature is 60oC.

(d) [4 marks] If TIN suddenly increases to 65 °C, whilst the air and steam temperatures remain
constant, sketch the time domain response of T.

Page 3 of 23
QUESTION 2 (10 marks):

A gas furnace is used to heat the feed to a reactor. The temperature T is controlled by a PID
controller, which adjusts the gas flow F. The gas supply pressure P tends to fluctuate. To overcome
this, cascade control has been implemented to regulate the gas flow supplied to the burners. The
inner controller is a proportional-only controller.

Process dynamics tests were conducted by the operators. When the waste gas pressure P increased
by 0.5 kPa, the gas flow F increased to 0.63m3/min after 1 minute, before reaching a new steady
state value of 1m3/min.

The change in exit temperature T due to a change in gas flow F can be modelled as a second order
system, with a gain of 3 oC/(m3/min), a time constant of 240 seconds, damping coefficient of 0.5.

When a unit step change is applied to the gas flow control valve, the gas flow F achieves a steady
state value of +5 m3/min/% after 5 minutes.

Both the temperature sensor and the gas flow sensor are very fast.

(a) [5 marks] Draw the block diagram of the system, showing all the transfer functions.

(b) [3 marks] Calculate the offset in F after a unit increase in the gas pressure P, if the inner
control loop is proportional only with a gain of Kc2=15 %open/(kg/m3).

(c) [2 marks] Determine the overall closed loop transfer function T(s)/SP(s).
(Note: you do not need to rearrange into the standard form).

Page 4 of 23
QUESTION 3 (15 marks):
A multiloop control system
consisting of 4 PID
controllers is to be designed
for a pH neutralisation
process. The transfer
function matrix and relative
gain array for this system
are shown below.

 0.43e −0.8 s 0.43e −0.1s 0.23e −1.0 s 0.22e −0.5 s 

 
4.32 s + 1 3.10 s + 1 5.24 s + 1 4.42 s + 1 
 h1   − 0.33e −1.0 s 0.32e −0.5 s − 0.20e −1.8 s 0.20e − 0.8 s  Q1 
 pH    
 1  =  2.56 s + 1 2.58s + 1 2.82 s + 1 3.30 s + 1  Q3 
 h2   0.22e −1.1s 0.23e −0.3 s 0.42e −0.4 s 0.41e −0.1s  Q4 
  

 pH 2   5.52 s +−11.5 s 4.49 s + 1 3.32 s + 1 2.07 s + 1  Q6 
 − 0.22e 0.22e −1.2 s − 0.32e −0.8 s 0.32e −0.4 s 
 3.24s + 1 2.65s + 1 2.36s + 1 2.03s + 1 

 0.64 0.72 − 0.20 − 0.20

 0.87 0.85 − 0.35 − 0.35
RGA= 
− 0.18 − 0.21 0.70 0.70 
 
− 0.36 − 0.37 0.85 0.88 

(a) [4 marks] Recommend a pairing of manipulated and controlled variables based on steady state
considerations.
(b) [3 marks] Would you also recommend this pairing based on physical and/or dynamic
considerations?
(c) [3 marks] Suppose that only pH2 and h2 are to be controlled using Q4 and Q6 as the manipulated
variables (Q1 and Q3 will be held constant)
i. What is the RGA for this 2x2 control problem?
ii. What pairing of controlled and manipulated variables do you recommend?

(d) [3 marks] Design two dynamic decouplers for the loop pairing you recommended in part (c)

(e) [2 marks] Draw a block diagram for the decouplers you designed in part (d), showing all the
transfer functions on the diagram.

Page 5 of 23
QUESTION 4 (15 marks):

Two stirred tank reactors are

connected in series. Some of the
product stream, CA2 (mol/m3) is
recycled back to the first tank as
shown in the diagram.

The transfer functions for this

system are given below.

C A1 ' 0.1 C A 2 ' 0.02 C A1 ' 0.04

= = =
C A0 ' 9 s + 1 C A1 ' 2s + 1 C A2 ' 9s + 1

(a) [5 marks] Draw the block diagram for this system.

(b) [5 marks] Determine the overall transfer function CA2’/CA0’.

(c) [3 marks] If that the overall response CA2’/CAO’ can be approximated by the following
TF:

C A 2 ' 0.002e −2 s
= where τ and θ are in minutes.
C A0 ' 10 s + 1

Calculate the dynamic matrix [A] for this process for a prediction horizon p=5 and a
control horizon c=3, with a sampling interval of 10 minutes.

(a) [2 marks] Determine [CA2] (where [CA2 ]=[A][CA0] ) for a step change in the inlet
concentration CA0 of 3 mol/m3 which occurs at t=10 min

Page 6 of 23
QUESTION 5 (15 marks):

(a) [5 marks] Explain how Dynamic Matrix Control works.

(b) [2 marks] Explain how DMC controllers can be tuned.

(c) [5 marks] Explain the action of (i) proportional, (ii) integral and (iii) derivative control.

(d) [3 marks] Briefly explain how cascade control works and when it is used.

END OF EXAM QUESTIONS

Page 7 of 23
 THIS PAGE HAS BEEN INTENTIONALLY LEFT BLANK

CHE3162 EQUATION SHEET

Y K
=
st
1 order: X τs + 1 ;

Page 8 of 23
 Y K
= 2 2
2nd order: X τ s + 2τξs + 1
 − πξ 
overshoot = exp  ; decay ratio = overshoot2
 1 − ξ2 
 
2πτ
T=
1− ξ 2
; tr=T/4 tp=T/2

FODT: τ = 1.5(t 63% − t 28% ); θ = t 63% − τ ; K=∆/δ

Time domain: y(t)=0 from t ≤0

  − (t − θ) 
y (t ) = K 1 − exp 
  τ  for t>0

K
AR =
(1 + τ 2ω 2 ) φ = tan −1 (− τω ) − θω

K1 K2 K3
AR =
1 + τ1 ω 2 1 + τ 2 ω2 1 + τ3 ω2
2 2 2

φ = tan −1 (− τ1ω) + tan −1 (− τ 2 ω) + tan −1 (− τ 3 ω)

 − 2ξτω  AR =
K
φ = tan −1  2 2 
1− τ ω  (1 − τ ω ) + ( 2ξτω) 2
2 2 2

Final Value Theorem: lim f(t) = sF(s)

t→∞ s→0

1
λ11 = , λ12 = 1 − λ11 = λ 21
K K
1 − 12 21
K11 K 22

Page 9 of 23
TUNING TABLES

Zeigler-Nichols Ultimate Sensitivity Tuning

Pu=2π/ωu
Kc TI τd
P 0.5Ku -- --
PI 0.45Ku Pu/1.2 --
PID 0.6Ku Pu/2 Pu/8

Cohen-Coon Tuning Rules:

R = θ/τ
Kc TI τd
(1 + R / 3)
P -- --
KR
(0.9 + R / 12) θ (30 + 3R)
PI --
KR (9 + 20 R)
(1.33 + R / 4) θ (32 + 6 R) 4θ
PID
KR (13 + 8 R) (11 + 2 R)

Laplace Transform Functions

f(t) f(s) = L [f(t)]

U(t), unit step or constant 1/s

t n −1 n!
(n − 1)! s n +1

(a − τ ) as + 1
1+ e − t /τ
τ s (τ s + 1)

τ 1 − a − t /τ τ 2 − a − t /τ as + 1
1+ e − 1
e 2

τ 2 − τ1 τ 2 − τ1 s (τ 1s + 1)(τ 2 s + 1)

1  1− ξ 2  1
1− e −ξ t /τ sin  t +φ 
1− ξ 2  τ


 s (τ s + 2ξτ s + 1)
2 2

 1− ξ 2 
φ = tan −1  
 ξ 
where  

Page 10 of 23
2012 EXAM SOLUTIONS

Page 11 of 23
Page 12 of 23
Page 13 of 23
Page 14 of 23
Page 15 of 23
Page 16 of 23
Question 3 solution
(a) Which is the best pairing?
(i) Static considerations (RGA): Pair according to RGA elements closest to +1:

 h1   0.64 0.72 − 0.20 − 0.20  Q1 

 pH   0.87
 1 =  0.85 − 0.35 − 0.35 Q3 
 h2  − 0.18 − 0.21 0.70 0.70  Q4 
    
 pH 2  − 0.36 − 0.37 0.85 0.88  Q6 
i.e. H1 – Q3, pH1-Q1, H2-Q4, pH2 – Q6
(4 marks)

Each row represents each output - so row 1 represents h1, and row 2 represents pH1, and row 3 is h2 and
row 4 is pH2, which is reasonably clear from the RGA. The columns also represent the inputs in order - so
column 1 represents Q1 and column 2 represents Q3 and column 3 is Q4 and column 4 is Q6.

So during the pairing, each output needs to be matched with only one input. This means that there can only
be one selection in each row and each column. If there are two numbers circled in any column or row then its
incorrect. This matrix is 4x4 but the top right 2x2 and the bottom left 2x2 are all negative numbers, so you
should narrow in very quickly on the only possible options.

0.85 and 0.88 are in the same row (last two columns of bottom row). The represent the options of controlling
pH2 with either Q4 or Q6. If both were selected, then we would be trying to control Ph2 with Q4 AND
Q6. So, you pick 0.88, which is largest and closest to one which means control pH2 with Q6, and that
means the 0.85 cannot be chosen because you have already decided the best stream to control Ph2 was
Q6. So the next highest value is 0.7, which is h2 and Q4.

Similarly, 0.87 in the 1st column, second row is the highest value for Q1. So you use Q1 to control
pH1. Then, this means the 0.85 in the next column cannot be used because you have already assigned pH1
to be controlled by Q1, and this column (2nd column) represents Q3. Since the 0.85 option is eliminated,
then the only other positive number os 0.72, which means control h1 with Q3.

(ii) Dynamic considerations: The some pairing results in the smallest time constants for tank 1 (see
st
highlighted TF Ph1/Q1 in the 1 column).
It is also dynamically best for tank 2 because it avoids the long deadtime of
1.8 min (see crossed out TF)
Physical considerations: The proposed pairing makes sense because the controlled variables for
each tank are paired with the inlet flows for that same tank.
Alternatively - The RGA recommends pH1 to be controlled by Q1, based on the gains, but looking at
the entire TF, this has a deadtime of 1 minute (highlighted TF). If you used Q3 to control pH1
instead then the deadtime is only 0.5min which might be better in practice.

 0.43e −0.8 s 0.43e −0.1s 0.23e −1.0 s 0.22e −0.5 s 

 
4.32 s + 1 3.10 s + 1 5.24 s + 1 4.42 s + 1 
 h1   − 0.33e −1.0 s 0.32e −0.5 s − 0.20e −1.8 s 0.20e − 0.8 s  Q1 
 pH    
 1  =  2.56 s + 1 2.58s + 1 X
2.82 s + 1 3.30 s + 1  Q3 
 h2   0.22e −1.1s 0.23e −0.3 s 0.42e −0.4 s 0.41e −0.1s  Q4 
  

 pH 2   5.52 s +−11.5 s 4.49 s + 1 3.32 s + 1 2.07 s + 1  Q6 
 − 0.22e 0.22e −1.2 s − 0.32e −0.8 s 0.32e −0.4 s 
 3.24s + 1 2.65s + 1 2.36s + 1 2.03s + 1 
[ 3marks for explaining any one of these, or any other reasonable answer]

Page 17 of 23
 b) Suppose that only pH2 and h2 are to be controlled using Q4 and Q6 as the manipulated variables (Q1
and Q3 will be held constant).
The new gain matrix for the 2⋅ 2 problem is

(1 mark)

( 1mark)
And the new RGA is :

RGA pairing: h2 – Q4 / pH2-Q6.

(1 mark – marking note: MUST list the pairing in words, not just show the RGA without explaining what it
means) (as a side note: this pairing also avoids the relatively large delay of 0.8 min. )

c) Decoupler design:
Pair H2 – Q4 and pH2-Q6.

h2 = G11Q4 + G12 Q6 pH 2 = G21Q4 + G22 Q6

0.42e −0.4 s 0.41e −0.1s − 0.32e −0.8 s 0.32e −0.4 s
h2 = Q4 + Q6 pH 2 = Q4 + Q6
3.32 s + 1 2.07 s + 1 2.36 s + 1 2.03s + 1
From the block diagram:
h2 = G11T12 Q6 + G12 Q6
h2
= 0 = G11T12 + G12
Q6
− G12
T12 =
G11
− 0.41e −0.1s 3.32 s + 1 − (3.32 s + 1)e + 0.3 s
T12 = =
2.07 s + 1 0.42e −0.4 s 2.07 s + 1
(1 mark)
− (3.32 s + 1)
This is not realisable, so ignore deadtime: T12 = ( 1mark for realisable)
2.07 s + 1

pH 2 = G21Q4 + G22T21Q4
pH 2
= 0 = G21 + G22T21
Q4
− G21
T21 =
G22
+ 0.32e −0.8 s 2.03s + 1 (2.03s + 1)e −0.4 s
T21 = =
2.36 s + 1 0.32e −0.4 s 2.36 s + 1
This is realisable so leave as is ( 1 mark)

Page 18 of 23
Page 19 of 23
Question 4

Page 20 of 23
Page 21 of 23
Page 22 of 23
QUESTION 5: Note – marking of this section requires significant marker judgement, and these
answers are a guide of the kind of things we are looking for. We mark each answer on its merits.

(a) [5 marks] Explain how Dynamic Matrix Control works.

For 3/5: Needs to explain multiple input and output control; basics of prediction horizon and
control horizon,
4/5: Might also give detailed example of how it works, mention tuning, quadratic equation
solution
(b) [2 marks] Explain how DMC controllers can be tuned.
• By adjusting adjust prediction and control horizons (1 mark)
• By adjusting Wu for the number of valve moves (1 mark)

(c) [5 marks] Explain the action of (i) proportional, (ii) integral and (iii) derivative control.
(i) Proportional control– takes action in proportion to the size of the error (or diff between
CV and SP). Simple and robust but produces Offset
(ii) Integral – looks at error between CV & SP over a time window and takes action to
eliminate this error. Removes offset but can add instability and oscillations
(iii) Derivative – reacts to rate of change of the error. Generally stabilises oscillations from PI
control although can make less stable. Can be sensitive to noise.

Basic answer – 1 mark each for P=proportional to error, I= eliminate offset by working on
sustained error (or error integral). And D = rate of change of error. Extra marks for more
information (but must be relevant! Not just a dump of memorised facts in the hope that at least
one of them will be correct!)

(i) [3 marks] Briefly explain how cascade control works and when it is used.
Cascade control is used when there is a disturbance to the manipulated variable (1 mark)
Adds a second inner loop around the valve to control the manipulated variable (1 mark)
Get 3/3 if mention these points plus something about how inner loop is usually high gain,
proportional only controller, or how inner loop needs to have faster response dynamics, or other
similar points.

Page 23 of 23