KMU 417 Process Control

A note about the use of these lecture notes
These lecture notes were prepared from the notes
of Prof. Dr. Erdoğan Alper by Dr. Eda Çelik-Akdur
and course assistant Özge Yüksel, also with the
help of class 2014-2015 section 22. The notes are
mostly based on the textbook by Stephanopoulos
(1984).
These are prepared only to aid class discussion.
Students are advised to use the textbook as there
are potentially some typing errors !
The notes may not be distributed to others without
permission or used out of the scope of this course.
KMU 417 Process Control
Hacettepe University
Department of Chemical Engineering
2014-2015 Fall Semester
Asst. Prof. Dr. Eda Çelik-Akdur
EdaCelik©
Textbook
KMU417 Course Goals

To comprehend and analyse the dynamic
behaviours of chemical process systems

To comprehend the process control techniques and
desing the chemical process control systems

To perform the simulations of chemical process
control systems with the help of MATLAB® and
SIMULINK®
EdaCelik©
3
2
Recommended
EdaCelik©
4
Attendance

Homeworks
•
Students have the right to attend 100% of the lectures.
MATLAB programs => the text of the code & output
on hard copy with the rest of your work. Otherwise, grade=0 !


If you miss more than 30% of the classes (12 hours) you
will get an F1 
•
If you are late for the first class, you may come in quietly, in
the first 5 minutes only ! But do not make this a habbit !
Please be on time for the second class hour.
•
Staple all pages of your work
•
Do not use a plastic envelope
•
Follow rules of ethics !
EdaCelik©
Late homework: will be accepted up to 2 days after the due date, but 20%/day.
5
EdaCelik©
Exams
6
Grading

2 midterm exams and a final exam

If you have to miss an exam due to medical reasons, you
may ask for a makeup exam, provided that you bring
proper documentation from your doctor.

Makeup exams will be much more challenging !

Follow rules of ethics !

Midterm exams
: 40 %

Homework and quizes
: 20 %

Final examination
: 40 %
* Full attendance
: 5% bonus
* Less then 70 % attendance: You may not take the final exam, F1 
* If your grade is less then 45 (out of 100) in the final exam, you get F3 
Remember that instructors don’t give the grades,
students earn the grades !
EdaCelik©
7
EdaCelik©
8
Incentives for Process Control
Control is necessary …
During its operation a chemical plant must satisfy
followings:
(1) Safety
(2) Product specifications
[ Production levels (tons/day), purity (%) ]
(3) Environmental regulations
(4) Operational constraints
(5) Economics
* These requirements necessitate continuous monitoring

to eliminate the effect of external
disturbances

to prevent runaway operation – stability

to optimise the performance
(measurement) of various parameters and adjustment
(control) of them.
EdaCelik©
9
EdaCelik©
10
Example 1.1 Stirred tank heater
Control steps
First: Measure
(measuring elements, sensors, such
as thermocouples, pressure or flow
transducers, etc.)
Second: Compare (with a set-point)
without control
‘‘Open Loop’’
Third: Action
(control using a final control element,
usually a control valve)
EdaCelik©
11
Fig. Ref: Stephanopoulos, 1984
with ‘‘Feedback’’ control
‘‘Closed Loop’’
Here Fi or Ti (inlet variables) may be
disturbed, then T is measured,
compared with its desired value and
steam rate is manipulated accordingly.
In the previous example, if we know the system
well, we may take action before the error happens !
Temperature (T) vs time
Consider, for instance; Fi = constant, i.e. it cannot
be disturbed. Then for a disturbence in Ti we may
propose:
‘‘Feedforward’’
Predictive Control
Fig. Ref: Erdoğan Alper, Lecturre notes
13
Fig. Ref: Stephanopoulos, 1984
14
What is Simulink good for?
Place of Process Control in a typical Chemical Plant
Ref: Luyben (1996)
EdaCelik©
15

Modeling/designing dynamic systems (including nonlinear dynamics)

Modeling/designing control systems (including nonlinear controllers)

Signal processing design/simulation
EdaCelik©
16
Simulink runs under Matlab. First
start Matlab, then type “simulink” at the
Matlab command window, or click on the
icon
EdaCelik©
17
EdaCelik©
18
A Simulink model is a block diagram. Click
“File|New|Model” in the Library Browser. An
empty block diagram will pop up. You can drag
blocks into the diagram from the library.
EdaCelik©
19
EdaCelik©
20
Sources: Produce Signals
EdaCelik©
Sinks: Terminate Signals
21
Connecting Blocks
EdaCelik©
Running the Simulation
* Drag a signal line from the
output of a block to the input
of another block.
Change parameters under
Simulation | Configuration
Parameters.
** Ctrl-Click will automatically
connect.
EdaCelik©
22
23
EdaCelik©
24
Running the Simulation
Running the Simulation
Once the
parameters are all
set,
click the play
button to run the
simulation.
EdaCelik©
25
EdaCelik©
26
Modifying Block Properties
Viewing Results: Scope
Double click on any block
to bring up a properties
box.
Here are the “sine wave”
properties.
If you don’t know
what something is…
leave it alone.
EdaCelik©
27
EdaCelik©
28
Adding Comments
Adding Signals or Signal Routing
You can create a branch point in
a signal line by holding down the
CTRL key, and clicking on the
line.
You can add text
comments anywhere
in the block diagram
by double clicking and
typing in some text.
A summer block (
) can be found in
the “commonly used blocks”
library, and in the “math” library.
You can change the
default comments
under the blocks by
double clicking and
editing the text.
EdaCelik©
To change the shape of the
summer to rectangular, or to add
additional inputs or change the
sign, double click on the summer.
29
EdaCelik©
30
CHP2: DESIGN APECTS of PROCESS CONTROL
Signal Routing
Classification of variables:
Under the “signal routing” library,
the MUX block can be used to
bundle a group of signals
together into a single line.
The DEMUX block does the
reverse.
Inputs
This can be useful to:
1. Clear up clutter in a
complicated block diagram.
2. Send multiple signals to the
same scope; then both
signals will be displayed on
the same plot.
EdaCelik©
Variables (flow rates, T, P, concentration)
31
Disturbances
 Measured
 Unmeasured
Outputs
Manipulated
(adjustable)
Measured
(controlled)
EdaCelik©
Unmeasured
32
Classificaton of Control Systems
Disturbance (d)
Single Input - Single Output (SISO) systems
Manipulated
variable (m)
PROCESSING
SYSTEM
(eg. a CSTR)
Measured
output (y)
Multiple Input – Multiple Output (MIMO) systems
(controlled
variable)
Unmeasured
output (z)
EdaCelik©
33
Fig Ref: Wikipedia
EdaCelik©
34
Design Elements of a Control System
Example 2.3. CSTR with a cooling jacket
(1) Define Control Objective: what are the operational
objectives of a control system (eliminate influence of
disturbance, stability, optimization, or combination of these)
(2) Select Measurements: what variables must be measured to
monitor the performance of a chemical plant (y, d, etc.)
(3) Select Manipulated Variables: consider alternatives (see Ex.
2.11)
(4) Select the Control Configuration: information structure for
measured and controlled variables. Configurations include:
(i) feedback control
(ii) feedforward control
(iii) inferential control
Disturbances (d) : CAİ, Fİ, Tİ, TC
Manipulated variable (m) : FC
Controlled variable (y) : T
Fig. Ref: Stephanopoulos, 1984
EdaCelik©
35
EdaCelik©
36
Control Configurations in a Distillation Column
How is the information taken from the measurements
used to adjust the values of the manipulated variables ?
Define Control Objective:
‘’Control law’’ which is implemented automatically by the
95 % top product
Select Measurements:
controller !
composition of Distillate
Select Manipulated
variables:
From the measurement of the controlled output and its
Reflux ratio
design value we obtain ‘‘error’’ or deviation (є) :
Select the Control
Configuration: feedback
control
y = value of the controlled output at any time t
ys = steady state (designed or desired) value of y
(Stephanopoulos, 1984)
37

We
Є = y – yS
EdaCelik©
38
Example 2.12: Stirred tank heater (Fig 1.1)
can also calculate :
i) Objective: To control T
ii) Measurement: Measure T by a
thermocouple
iii) Manipulated variable:
Heat input by steam, Q
iv) Configuration: Feedback
How do we use this information for manipulation ?
control
How should Q change in order to keep T constant,
when Ti is disturbed ?
EdaCelik©
39
EdaCelik©
40
Subtract


V ρ Cp
EdaCelik©
41
= F ρ Cp T i − T + Q
(2.3)
We want to drive this error (є = T ‘) to zero, by manipulating Q,
How ?
EdaCelik©
42
We may make an adjustment in Q in proportion to
the error (є) :
Q ’ = Q – Qs = - α ( T- Ts )
(2.4)
This is called ‘‘Proportional Control’’ (P-control), where
(Stephanopoulos, 1984)
- α (= Kc) = proportional gain
Important Conclusion: In P-control, increasing
Combine Eqn (2.3) and (2.4) :
V ρ Cp
= F ρ Cp T i − T
− α T’
P-gain ( α ) decreases ‘‘offset’’ (i.e. final error)
(2.5)
but we can not eliminate it completely.
Eqn. (2.5) is a linear initial value ODE which can be solved easily
EdaCelik©
43
EdaCelik©
44
Con’t, try integral control:
V ρ Cp
= F ρ Cp T i − T
− α’∫
Alternatively, use proportional-integral (PI) control,
then value of heat inout (Q) is given by :
′
Q’ = - α T’ – α’∫
Solution for different α’ yields the graph below
where,
(Ref: Stephanopoulos, 1984)
EdaCelik©
45
′
α
= τI = Integral time
α
(Ref: Stephanopoulos, 1984)
EdaCelik©
46
Chp 3: Hardware for a Process Control System







DDC: Direct Digital Control
The process (chemical or physical)
Measuring instruments and sensors (inputs, outputs)
what are the sensors for measuring T, P, F, h, x, etc?
Transducers (converts measurements to current/
voltage, eg: a strain gauge)
Transmission lines (usaully, electric signals)
The controller (with machine ‘‘intelligence’’)
The final control element (eg: a control valve, variablespeed pumps, etc.)
Recording/ display
elements
Recall Process
Instrumentation
Fig. Ref: http://www.ddc-online.org/
(Stephanopoulos, 1984)
47
EdaCelik©
48
Primary controllers typically have the
following features:









Summary of Part I (Chps 1-3)
Real-time accurate clock function
Full software compliment
Larger total point capacity
Support for global strategies
Buffer for alarms/messages/trend & runtime data
Freeform programming
Downloadable database
Higher analog/digital converter resolution
Built-in communication interface for PC
connection.
EdaCelik©
49
 Incentives
 Elements
for process control
of process control and control
laws
 Hardware
elements in basic control
EdaCelik©
50