ChE 441

Process Control
Fall 2014

SRSaunders - WSU - ChE 441

Download Socrative
Clicker Replacement
Not

just multiple choice.

Sometimes anonymous, sometimes not

Does

not impact grade

Provides feedback

Apps available for iPhone/iPad/Android

Access via web http://b.socrative.com
Class ID: CHE441

SRSaunders - WSU - ChE 441

Syllabus General 1
Instructor
Dr.

Steven R. Saunders
Office: Wegner 213
Phone: 5-6578
Email: steven.r.saunders@wsu.edu

Office Hours:
TBD
First

Socrative question!

Discuss with two people next to you: What are 3

good times for office hours?
SRSaunders - WSU - ChE 441

Syllabus General 2
Lectures
MWF

9:10am 10:00am
G1 Wegner

Prerequisites (C or better)
ChE

211 (Process Simulation)

ChE 310 (Intro to Transport Processes)

Course Description
Measuring

instruments, automatic control, process and

instrument characteristics and theory applied to industrial
control problems.
3 credits

Communication via Email

Emails

will only be sent to @email.wsu.edu or

@wsu.edu
Ensure forwarding is properly set up.
SRSaunders - WSU - ChE 441

Syllabus - Textbooks
Process Dynamics and Control (Required)
Seborg, Edgar, Mellichamp, and Doyle
3rd Edition
Wiley & Sons, 2014.
Custom Version: 978-1-11894-672-5
Full Version: 978-0-470-12867-1

Addition Problem Sets from (not required)

Process Dynamics, Modeling and Control
Babatunde A. Ogunnaike and W. Harmon Ray

Principles and Practice of Automatic Process Control

Smith and Corripio

Chemical and Bio-Process Control

Riggs and Kazim

Process Control
Bequette

SRSaunders - WSU - ChE 441

Syllabus - Webpage
Webpage maintained on Angel
Relevant

documents will be posted

Student versions of lecture notes
Full versions will not be distributed unless you have a
valid university/medical excuse
Discussion

forum

Please post questions prior to coming to office hours

If you have a question, someone else probably does
too
You can post anonymously
If you answer the question (correctly) it will have an
impact on your grade.
SRSaunders - WSU - ChE 441

Syllabus - Homework
Due at beginning of class
Late homework not accepted w/o official excuse
Of your own work
Homework must meet the official course
HOMEWORK FORMAT
You must obtain a minimum of 50% on
homework assignments in order to make a C
or better in this class regardless of your
grades on the quizzes, hour exams, and the
final exam. (i.e., this requirement stands
alone).
SRSaunders - WSU - ChE 441

Syllabus - Quizzes
Announced and unannounced
No make-ups w/o an official excuse
Given at the beginning of class
NO

EXTRA TIME FOR LATE ARRIVALS

You must obtain a minimum of 50% on

the quizzes in order to make a C or
better in this class regardless of your
grades on the homework, hour exams,
and the final exam. (i.e., this
requirement stands alone).
SRSaunders - WSU - ChE 441

Syllabus - Exams
TWO mid-term exams
No extra time for late arrivals
No make-ups w/o official excuse
Final Exam
2

hour comprehensive exam

December 19th, 2013
8:00am 10:00am
No extra time for late arrivals
Requests for special accommodations will only be
considered prior to the student taking any exam.
Once the student has taken an exam, no re-tests
will be given.
SRSaunders - WSU - ChE 441

Syllabus - Grading
Grading
Exam

1
25%
Exam 2
25%
Homework
12.5%
Quizzes
12.5%
Final Exam
25%
Total
100%
Scor
e

9
0

<90
86

Gra
de

A-

<86
84
Aor
B+

<84
82

<82
78

<78
76

B+

B-

SRSaunders - WSU - ChE 441

<76
74
Bor
C+

<74
72

<72
68

<68

C+

10

Syllabus - Grading
Numerical grades will not be rounded.
If a numerical grade falls into one of the gray areas ( Aor B+, B- or C+), the letter grade will be determined by
overall performance trends, by consistency in
homework, and by final exam performance. There is no
gray area between any other letter grades (including
between a C and an F).
An unofficial grade book will be maintained on the
course webpage.
Re-grading of any graded material must be done within
2 weeks of the material being returned.
I reserve the right to re-administer any test or quiz. No
grades will be deleted. The grade for the readministered test of quiz will be used to supplement the
original grade. Specifics will be announced if necessary.
SRSaunders - WSU - ChE 441

11

Syllabus - Grading
A/A The student clearly demonstrates an in-depth technical
understanding of the concepts
B+/B/B The student demonstrates a technical understanding sufficient
for solving the majority of problems
C+/C
The student demonstrates a minimal technical understanding
sufficient for solving straightforward problems but may have
trouble with more complex variations or situations
C-/D+/D
These will not be given
F
The student's ability to apply the concepts to problems is
seriously in question.

SRSaunders - WSU - ChE 441

12

Syllabus Academic Honesty

Academic integrity will be strongly enforced in
this course. Any student caught cheating on
any assignment will be given an F grade for
the course and will be reported to the Office
Student Standards and Accountability.
Cheating is defined in the Standards for
Student Conduct WAC 504-26-010 (3). It is
strongly suggested that you read and
understand these definitions:
http://apps.leg.wa.gov/wac/default.aspx?
cite=504-26-010
SRSaunders - WSU - ChE 441

13

Syllabus
Special Accommodations
Reasonable

accommodations are available for students with a

documented disability. If you have a disability and may need
accommodations to fully participate in this class, please visit the
Disability Resource Center (DRC). All accommodations MUST be
approved through the DRC (2nd Floor, Washington Building).
Please stop by or call 509-335-3417 to make an appointment
with a disability specialist.

Campus Safety
All

students should be familiar with the Campus Safety Plan and

the University Emergency Management system. Details on these
can be found at http://safetyplan.wsu.edu and
http://oem.wsu.edu/emergencies, respectively. Everyone is
encouraged to go to http://my.wsu.edu and, under the
Emergency Notification box, enter their emergency contact
information.

SRSaunders - WSU - ChE 441

14

Syllabus
Tentative Course Schedule (subject to
change)

9/1/2014 No Class (Labor Day)

9/22/2014 Last day to Drop Course
9/29/2014 End of Material for Exam 1
10/3/2014 Exam 1
10/15/2014 Mid-term grades submitted
11/3/2014 End of Material for Exam 2
11/7/2014 Exam 2
11/17/2014 -> 11/19/2014 Saunders @ AIChE
Week of 11/24/2014 No Class (Thanksgiving)
12/12/2014 Last day of instruction
12/19/2014 Final Exam
SRSaunders - WSU - ChE 441

15

Why Study Process Control

Academic Motivation
Integrate

knowledge from previous courses

Improve/solidify math and computational
skills
Learn systems thinking

Exercise:
What

is the general form for a mass or energy

balance?

Accumulation

= In Out + Generation

SRSaunders - WSU - ChE 441

16

Industrial Motivation for Process

Control

Necessary for
Safe

operation
Reliable (consistent) operation
Efficiency of operation
Product quality

Major impact on profitability

Career Option: Process Control Engineer
High

Mobility
High Visibility
Technically Challenging
SRSaunders - WSU - ChE 441

17

A refinery

SRSaunders - WSU - ChE 441

18

Another Example
See Loop-Pro Control

SRSaunders - WSU - ChE 441

19

What will we be doing?

What do you think are the steps to develop
an effect control system?

SRSaunders - WSU - ChE 441

20

Nomenclature
Variable something that changes with time
Input (variable) independently changes the
process operating state (Fi, cAo, R, Q)
Manipulated

(u)
Disturbance (d)

Output (variable) provides information about

process
Measured

sensors provide info (level, GC,

conductivity)
Controlled (y)
Uncontrolled

Unmeasured

no measurement device available

Setpoint
SRSaunders - WSU - ChE 441

21

Example Driving a Car

Control Objective (setpoint) Speed Limit
(70 mph)
Controlled Variable Speed
Manipulated Variable- Position of
Accelerator

SRSaunders - WSU - ChE 441

22

Elements of a Control System

Process distillation column, holding tank,
etc.
Sensor measuring devices
Controller decision makers (why we are
here)
Transmitter communicates information
from sensor to controller
Final Control Element valves, pumps,
compressor, switches

SRSaunders - WSU - ChE 441

23

Example Driving a Car

Control Objective (setpoint)
Controlled Variable
Manipulated Variable Sensor
Controller
Final Control Element Disturbance SRSaunders - WSU - ChE 441

24

Types of Control
Regulatory
Control

system is only responsible for

counteracting disturbances

Servo
Responsible

for adjusting the set-point

SRSaunders - WSU - ChE 441

25

Example.. Servo Vs Regulatory

Take 30 seconds How does a sprinkler
system work?

SRSaunders - WSU - ChE 441

26

Example Shell and Tube Heat

Exchange

What are the control

objectives?
What would be
manipulated to
achieve this
objective?
Are there any
disturbances that
could affect the
system?

SRSaunders - WSU - ChE 441

27

Example Heat Exchanger

Control Objective:

Stea
m

Controlled Variable

Feed
Manipulated Variable T
i

Sensor

Final Control Element

Disturbances?
SRSaunders - WSU - ChE 441

Condens
ate

Outle
t
To

Control Schemes No Control

Disturbance

Input

Output

Process
SRSaunders - WSU - ChE 441

29

Control Schemes Open Loop

Disturbance

Input

Output

Process
Transmitter
Final
Control
Element

Controller
SRSaunders - WSU - ChE 441

30

Simple Heat Exchanger Open Loop

Stea
m

Feed
Ti

Outl
et
To

Condensa
SRSaunders - WSU - ChE 441
te

31

Control Schemes Feedback

Measure controlled variable
Adjust manipulated variable to keep
controlled variable at setpoint

Disturbance

Input

Output

Process
Final
Control
Element

Transmitter

Sensor

Controller

SRSaunders - WSU - ChE 441

32

Simple Heat Exchanger Feedback

Steam

TC

Feed
Ti

Outlet
To

TS

Condensate
SRSaunders - WSU - ChE 441

33

Control Schemes Feedforward

Measure disturbance variable
Adjust manipulated variable to compensate
for disturbance

Disturbance

Controller

Sensor
Transmitter

Input

Output

Process
Final
Control
Element

SRSaunders - WSU - ChE 441

34

Simple Heat Exchanger

Feedforward
Steam

TC
Feed
Ti

Outlet
To

TS

Condensate
SRSaunders - WSU - ChE 441

35

Controller Comparison
Feedback
(+) Simple
(+)
(+) Use output
measurement to
determine control
(+)

(-) Disturbances must be

manifest in output before
action can be taken
SRSaunders - WSU - ChE 441

Feedforward
(+)
(+) Manipulated
variable changes before
disturbance manifests
in output
(-)
(-)

36

Nomenclature
Variable something that changes with time
Input (variable) independently changes the
process operating state (Fi, cAo, R, Q)
Manipulated

(u) at the discretion of the controller

Disturbance (d) random fluctuations

Output (variable) provides information about

process
Measured

sensors provide info (level, GC,

conductivity)
Controlled (y) relevant to process performance
Uncontrolled

Unmeasured

no measurement device available

Setpoint desired value of the controlled variable

SRSaunders - WSU - ChE 441

37

Block Diagrams
General method for approaching systems
Visualization of process behavior
One block for each element of process
Contains both physical and information flow
Open Loop Manual manipulated variable
change induces change in output

Closed Loop Controller automatically adjusts

manipulated variable to account for changes
SRSaunders - WSU - ChE 441

38

General Rules for Block Diagrams

Blocks represent mathematical models
Each block has 1 input and 1 output
Streams represent information flow and not
material flow
Split streams are copies not actually splits
Streams can be added or subtracted to
another but not multiplied or divided by
another

SRSaunders - WSU - ChE 441

39

Heat Exchanger Block Diagram

Steam
Outlet
To

Feed
Ti

Steam Feed
Pressure

Condensate

Feed
Ti

Valve
Position

SRSaunders - WSU - ChE 441

Outlet
To
40

Feedback Block Diagram

Steam

TC

Feed
Ti

TS

Outlet
To
Steam Feed
Pressure

Feed
Ti

Condensate

Desired Temp
(Setpoint)

SRSaunders - WSU - ChE 441

Outlet
To

41

Feedforward Block Diagram

Steam

Feed
Ti

TC

Outlet
To

TS

Feed
Ti

Condensate

Outlet
To

SRSaunders - WSU - ChE 441

42

Block Diagram Summary

Clean method for approaching and visualizing
process (information/material) flow
Each block contains a mathematical model
Convenient

model structure facilitates

simulation and solutions

Things to think about

How

are models constructed?

What is a valid model?
What complications are possible
How to handle block diagram math

SRSaunders - WSU - ChE 441

43

Process Modeling Motivation

Why model a process?
Improves

process understanding
Surrogate for physical system
Rapid and inexpensive to test process dynamics
Implies that the model is predictive (extrapolative)
Facilitates

controller synthesis

Assumptions
All

models are initially (t0) at a steady-state

At t0 a change is made to a manipulated input
Model needs to capture t >0

WARNING
Models

are NOT real-world

Models cannot capture ALL process behavior
Models are only as accurate as the equations and the author

SRSaunders - WSU - ChE 441

44

Example Tank Level Control

Problem:
A

cylindrical tank having a cross sectional area of

2 m2 provides the feed stream to a distillation
column. The steady-state feed rate to the tank is
2m3/min and the tank drains proportionally to the
height of liquid (proportionality constant c=2). It
is desired that the tank neither overflow or run
dry. Develop a model for the height of liquid in
the tank as a function of time and feed flowrate.
The fluid can be assumed to be constant density

Control Objective
Control

the liquid level in the tank

SRSaunders - WSU - ChE 441

45

Example Tank Level Control

Fin

Fout

One possible solution:

SRSaunders - WSU - ChE 441

46

Block Diagram for the Tank

Open-Loop

Engineer

Feedback Control

Fin

LC

LS

Fout

SRSaunders - WSU - ChE 441

47

Example Stirred Heated Tank

Problem
Liquid

at a temp. or 80C (Ti) flows into the tank at

a volumetric flowrate of 20 kg/min (win). The liquid
is heated by a heating element at a rate of 200
J/min (Q). The heated fluid, now temp. T, is
withdrawn at the same flowrate as the inlet stream
(wout = win). The tank volume is 5 m3 (V), the liquid
density is 2 kg/m3 (), and the specific heat
capacity is 5 J kg-1 C-1 (Cp).
Sketch the process and instrumentation diagram
Develop the dynamic energy balance
Sketch the open look process diagam
What is T at t= ?

SRSaunders - WSU - ChE 441

48

Stirred Heating Tank Assumptions

Contents of the tank are well mixed
Physical properties do not vary with
temperature
Adiabatic (i.e., heat losses to the
environment are negligible)

SRSaunders - WSU - ChE 441

49

Example Stirred Heating Tank

win
Ti

wout

Q
Ti
Q

Gd

Tank SRSaunders - WSU - ChE 441

50

Example Stirred Heating Tank

The plant manager wants the tanks exit
stream to have a final temperature of
110C
Identify:

manipulated variable, controlled

variable, measured output, and disturbance
variable
Incorporate a feedback controller (include the
sensor) into your process sketch
Construct the closed loop block diagram
Label all blocks and signals

SRSaunders - WSU - ChE 441

51

Example Stirred Heating Tank

Fin
Manipulated variable:
Ti
Controlled Variable:
Measured Output:
Disturbance Variable:

Fout

Thermocouple
SRSaunders - WSU - ChE 441

52

Exercise 1st order RXN in CSTR

The 1st order reaction A -> B, with a rate constant k = 1.8
min-1, occurs in a CSTR V = 100L. The feed flowrate is F in
= 20 L/min and the concentration of A in the feed is c Af 10
g/L.
Sketch

the open-loop process

Develop a model for the process (dynamic mass balance)

The plant manager wants to sell the reactors exit stream

with a concentration of cB = 9 g/L
Identify

the manipulated variable, controlled variable,

measured variable, and disturbance variables
Incorporate a feedback controller (including sensor) into your
process sketch
Construct the closed loop block diagram, label all blocks and
signals

SRSaunders - WSU - ChE 441

53

Exercise 1st order RXN in CSTR

Fin
cAf

cA
V

Fout
cA

SRSaunders - WSU - ChE 441

54

Exercise 1st order RXN in CSTR

Manipulated Variable ->

Fin

Controlled Variable ->

cAf

Measured Variable ->

Disturbance Variables ->

c A, c B

Fout

SRSaunders - WSU - ChE 441

c A, c B

55

Solving Linear ODEs

Standard Form 1st Order Linear ODE

General Solution

Particular Solution
Use

initial conditions

SRSaunders - WSU - ChE 441

56

Example
Solve
Q

SRSaunders - WSU - ChE 441

57

What makes an ODE Linear

General 1st order ODE

Linear if and only if

Only

time the general solution is valid

Special case,

Solvable

by separation of variables

SRSaunders - WSU - ChE 441

58

Flashback Liquid Level in a Tank

Fin

Fout

Solve that ODE!

SRSaunders - WSU - ChE 441

59

Flashback Liquid Level in a Tank

SRSaunders - WSU - ChE 441

60

Flashback Liquid Level in a Tank

Problem: Find the particular solution, and
explain it physically, and does it make
sense given the problem statement?

SRSaunders - WSU - ChE 441

61

Solving 2nd Order ODEs

General Form of 2nd Order ODEs

We will only deal with homogeneous 2nd

order ODEs
f(t)

=0

General Solution

SRSaunders - WSU - ChE 441

62

Solving 2nd Order Homogeneous ODEs

Given
Use Characteristic Equation
Roots of Characteristic Equation describe
g(t) and h(t)
Two

distinct REAL roots

Two equal REAL Roots
Two COMPLEX Roots

SRSaunders - WSU - ChE 441

63

Solving 2nd Order Homogeneous ODEs

Two

distinct REAL roots (m1m2)

Two

equal REAL Roots (m1 = m2)

Two

COMPLEX Roots (m1 = +i, m2 = -i)

SRSaunders - WSU - ChE 441

64

Exercise Solve That ODE

Given

, solve for y

Use Characteristic Eqn.

Solve Characteristic Eqn.
using quadratic formula
Solve quadratic
Two COMPLEX Roots

SRSaunders - WSU - ChE 441

65

Real Tank Problem

Control Objective:
Control

Fin

LC

the liquid level in the tank

LS

Tank exit flowrate from bottom mounted orifice is

proportional to the SQUARE ROOT of the liquid level
But we can only solve linear equations
How

do we address this issue of nonlinearity?

SRSaunders - WSU - ChE 441

66

Fout

Nonlinear Systems
What is an example of a nonlinear system?
Examples

of nonlinearity

Tools we develop are only for linear

systems
Solution
Find

a linear approximation

Method
Taylor

Series Expansion

SRSaunders - WSU - ChE 441

67

Taylor Series Approximations

Taylor Series

Steady State value of x is xs

Linearization truncates after linear term

Linearize f(x) = x2

SRSaunders - WSU - ChE 441

68

Local Linearization
30
25

20
15
10
5
-5

-4

-3

-2

0
-1
0
-5

-10

Linearization only valid locally, NOT globally!

SRSaunders - WSU - ChE 441

69

The nonlinear Tank Problem

Control objective: control the level
Your objective: Construct a linear model of
the tank
LC

Fin

LS

SRSaunders - WSU - ChE 441

70

Fout

General Linearization of ODEs

Nonlinear ODE

Linearize only nonlinear terms

Linearization of linear terms returns the
same linear term
SRSaunders - WSU - ChE 441

71

Exercise 2nd Order RXN in CSTR

The second order reaction A->B occurs in a
CSTR. Construct the nonlinear process
model as well as the local linear
approximation.

SRSaunders - WSU - ChE 441

72

Deviation Variables
Normal process operating condition is
nominal
s in xs denotes steady state
We will be considering changes to steady
state
Convenient to examine deviation of
process from its nominal condition
Deviation variables:
Controlled

Output
Manipulated Input
SRSaunders - WSU - ChE 441

73

Deviation Variables - Tank

Tank
Tank @ SS
Deviation

Rewrite

SRSaunders - WSU - ChE 441

74

Deviation Variable - Tank

Objective: Find y(t) from

When u(t) is
Constant
Ramp
random
Etc.

Integrating the ODE isnt too hard but very tedious for
different u(t)
Use Laplace transform!

SRSaunders - WSU - ChE 441

75

Laplace Transforms
Tool for solving linear ordinary
differential eqns.
Has

lots of neat properties

Defintion: Laplace transform of function

f(t)

Note: Dropping the ^ from here on out

SRSaunders - WSU - ChE 441

76

Two examples
Find the Laplace transform of

Find the Laplace transform of

SRSaunders - WSU - ChE 441

77

Laplace Transform Properties

operator converts from t to s (complex
variable)
Casual Operator
F(s)

has no information for

Good for control problems: nothing can be
done about the past, only for now and the
future

is a linear operator
Inverse Laplace is unique
SRSaunders - WSU - ChE 441

78

Lucky for you

Tables exist for {f(t)} and for -1{F(s)}
Official table for ChE 441
You will be able to use it MOST of the time
i.e.,

NOT ALL OF THE TIME

SRSaunders - WSU - ChE 441

79

General Problem Solving (1)

SRSaunders - WSU - ChE 441

80

Back to the Real Tank Problem

Fin

LC

LS

Solve by Laplace Transform (no more

integrating)
SRSaunders - WSU - ChE 441

81

Fout

Solving the Tank Problem

Tank Model

Take the Laplace Transform

Use Linearity Property

Laplace Definition

SRSaunders - WSU - ChE 441

82

of a Derivative?

Definition

SRSaunders - WSU - ChE 441

83

More of a Derivative
Higher Order Derivatives

is the (n-1)th derivative of f evaluated at 0

Example: Calculate

SRSaunders - WSU - ChE 441

84

Even More of a Derivative

Why do we use deviation variables?
At

t = 0, we assume our process is already at

steady state

Laplace Transform of integrals

SRSaunders - WSU - ChE 441

85

Solving the Tank Problem

What is y(0)?
Find Y(s) as a function of U(s)

SRSaunders - WSU - ChE 441

86

Laplace Transform Practice

The second order reaction A->B occurs in a
CSTR. Assume F is constant Recall the
linearized form:

Find Y(s) as a function of U(s)

Convert

into deviation variables

Take the Laplace transform
Solve

SRSaunders - WSU - ChE 441

87

Laplace Practice (Workspace)

SRSaunders - WSU - ChE 441

88

Laplace Practice (Results)

Most 1st order ODEs get

to something that looks
like this.

SRSaunders - WSU - ChE 441

89

Determining Values of y(t)

Find
y(t), y(0), and y() for the reactor

Need to know U(s) to evaluate

Once

you know U(s), use tables to determine y(t)

But we dont need to evaluate to determine

y(0) and y()
SRSaunders - WSU - ChE 441

90

Value Theorems
Final Value Theorem
Requires

f() <

Initial Value Theorem

Suppose u(t) = 1
Find y(0), and y() for

SRSaunders - WSU - ChE 441

91

Evaluating Value Theorems

SRSaunders - WSU - ChE 441

92

General Laplace Transformation

Construct Process Model
Linearize
Place in deviation variables
Convert time-domain variables into
Laplace domain
Solve algebraic equations for Y(s)
Calculate inverse Laplace (using tables)
Result: y(t)

SRSaunders - WSU - ChE 441

93

Determining Inverse Laplace

This is in the table

but what if it wasnt?

Use Partial Fractions!

SRSaunders - WSU - ChE 441

94

Using Partial Fractions

Use Partial Fractions
Multiply both sides by the
denominator on the left
Use test cases, pick values for s
to solve for A/B
When

When

SRSaunders - WSU - ChE 441

95

Evaluating Inverse Laplace

Use Partial Fractions

Use Linearity
Look at the table and
do maths!
Use table!

SRSaunders - WSU - ChE 441

96

Another Partial Fractions Example

Find y(t) for the process and input modeled
as:

SRSaunders - WSU - ChE 441

97

Another Inverse Laplace Practice

SRSaunders - WSU - ChE 441

98

Partial Fractions Expansion

Headaches

Dealing with repeated roots

What about A? Pick any number

SRSaunders - WSU - ChE 441

99

Partial Fraction Headaches

SRSaunders - WSU - ChE 441

100

More Partial Fractions Headache

Extra credit: Find y(t)

SRSaunders - WSU - ChE 441

101

Time Delay
Time delay exists and must be addressed
Example How do you think concentration
is typically measured?
GC

can take 15 minutes

How can our model handle an output

Example How long does it take for a

change in concentration in the feed to
effect the tank if there is a 50 yard tube in
between the tank and control?

SRSaunders - WSU - ChE 441

102

Accounting for Time Delays

Let is take time units for a change in cAf
to effect the reactor

The SHIFT PROPERTY

Traditional
Definitions
not the most
applicable
SRSaunders - WSU - ChE 441

103

Using the Shift Property

A modified reactor problem

SRSaunders - WSU - ChE 441

104

Time Delay Systems

Physical Example
How

Long until dye injected @ A reaches

BDye
?

v
A

So

yB(t) is identical to yA(t-)

Use

the shift in time property

SRSaunders - WSU - ChE 441

105

Time Delay Approximations

Time delay (as e-s) can be inconvenient
Analysis

and stability
Understanding
Mathematically (e-s can show up in the
denominator and cause problems)
Controller design
Pad

Approximation

SRSaunders - WSU - ChE 441

106

Forcing Function
Controlled variable changes result from
changes in process inputs
Manipulated

variables (u(t))
Disturbance variables (d(t))

Dynamic analysis uses well characterized

and well defined inputs
Step
Rectangular

pulse (2 steps in series)

Impulse
Ramp
Sinusoid

SRSaunders - WSU - ChE 441

107

The Step Function

Most
common forcing function

Ideal step of magnitude A

is the Heaviside Function (also called the

unit step function)
SRSaunders - WSU - ChE 441

108

u(t)

u(t)

The Heaviside Function

0
0

0
t

SRSaunders - WSU - ChE 441

b
109

More Step Functions

u(t)

-2 -1

time

What is the value of A in the Figure?

SRSaunders - WSU - ChE 441

110

Rectangular Pulse
Pulse of Magnitude A, and duration b

4
3

A=?

2
u(t) 1

b=?

0
-1

-2

-1

time

SRSaunders - WSU - ChE 441

111

Building a Rectangular Pulse

Remember we always work in deviation
variables, always from the initial steady
state..
Composed of two step functions
Based on the last slide, sketch the two
steps
0
-1
u(t)

-2
-3
-4

-2

-1

time
SRSaunders - WSU - ChE 441

112

Building a Pulse

SRSaunders - WSU - ChE 441

113

The Impulse Function

Im(PULSE)
height
Zero width
Area = A

u(t)

Infinite

0
0
SRSaunders - WSU - ChE 441

114

Step and Pulse Notes

Important Properties
Unit

Pulse (t) (Dirac Delta): Special case of

rectangular pulse. A x b = 1, as A -> b -> 0

Impulse

is related to step by derivative

Steps, pulses, and impulses are ideal

function and cannot be realized exactly
But

we can approximate them

SRSaunders - WSU - ChE 441

115

The Ramp Function

Ramp with slope A
31
26
21
16
u(t)

11
6
1
-4
-2

-1

time

SRSaunders - WSU - ChE 441

116

Lets talk about a delayed ramp

Ramp with no delay

Ramp with delay

Lets take the inverse

Laplace using the
Traditional Definition

Is this correct?
NO!
SRSaunders - WSU - ChE 441

117

Lets talk about a ramp with a delay

14
12
10
8
6

u(t)

That doesnt look

right it should be
zero until t=4 but it
isnt

4
2
0
-2
-4
-6
0

10

10

time
14

It should be

12
10
8
6

u(t)

4
2
0
-2
-4
-6
0

time

SRSaunders - WSU - ChE 441

118

Last one on delayed ramps

Correct Inverse Laplace of a delayed
system
Lets look at the ramp is really

So a delayed ramp is

SRSaunders - WSU - ChE 441

119

DELAYS... ONE SLIDE TO RULE THEM

ALL

To take the Laplace of a delayed function

To take the inverse Laplace of a delayed

function

Why? We need to make sure we tell responses to start

deviating from 0. Accounting for the delay requires two
modifications: 1) a shift in the function and 2) some
single that says dont act too early.
What if = 0?
SRSaunders - WSU - ChE 441

120

A Delayed response
First order system delayed step input
Inverse Laplace neglecting the Heaviside
Inverse Laplace with the Heaviside
4
3
2
1

y(t)

0
-1

-2

Reponse BEFORE the

step even occurs
SRSaunders - WSU - ChE 441

-3
-4
-5

time

121

10

The Sinusoid
Sine (or cosine)
Amplitude

Period

2
1
u(t)

0
-1
-2
-3

-2

-1

time

SRSaunders - WSU - ChE 441

122

Practice
Sketch the time domain input (label values
on the axes), and compute u(t)

3
1
u(t)

-1
-3

-1

7
time

11

13

15
123

Practice (results)
4
3
2
1
u(t)

0
-1
-2
-3

-1

11

13

time

SRSaunders - WSU - ChE 441

124

One more example

Find U(s) and u(t) for the following input
signal
11
9
7
5
u(t)

3
1
-1
-3

-1 0 1 2 3 4 5 6 7 8 9 10
time

SRSaunders - WSU - ChE 441

125

The Transfer Function

Inputs affect the output
Transfer function transfers input effect
into output response

Fin

Fd

Fd

Tank

Fin

Fout

Tank

Disturbance

Manipulated

SRSaunders - WSU - ChE 441

126

SRSaunders - WSU - ChE 441

127

Transfer Function Poles

General Form

Pole root of the denominator,

Defines

system stability

Identifies

periodic behavior (oscillation)

SRSaunders - WSU - ChE 441

128

Transfer Function Zeroes

Zero roots of the numerator,
Does

not alter exponents in y(t)

Does alter leading coefficient

Can

cancel poles

SRSaunders - WSU - ChE 441

129

Transfer Function Characteristics

Order ( )||
Highest

power of NUM(s) or DEN(S)

Overall transfer function order given by o(DEN(s))

Physical realizability (can we build a device)

Strictly

proper

Proper
Improper

Standard form
so

coefficient of NUM(s) and DEN(s) = 1

SRSaunders - WSU - ChE 441

130

Transfer Function Practice

Identify the number and value of poles and
zeros, and the order for:

SRSaunders - WSU - ChE 441

131

Half Planes
Remember s is a complex number
Can plot in complex space
Real component defines half-planes
Im
LHP

RHP

Re

Where does Re(s)=0 fall?

SRSaunders - WSU - ChE 441

132

Stability
Stability is defined by pole location
Right

Half Plane pole -> UNSTABLE

Exponent is positive
What happens as t->?
Left

Half Plane pole -> Stable

SRSaunders - WSU - ChE 441

133

Stability and Half Plane Practice

Identify the poles
and zeroes
Is the system
stable?

4
3

Pole
s

HW Show ALL
1
steps:What is G(s) if
0
-4 -3 -2 -1 0
the gain (K) is 3?
-1

-2
-3
-4
SRSaunders - WSU - ChE 441

134

Extra credit
Draw the Pole-Zero Complex plane for the
follow transfer function and is it stable?

SRSaunders - WSU - ChE 441

135

Properties of Linear Systems

Superposition
Symmetric Responses
Step

up of A give the opposite but exact

response as a step down of A

Scaling
Doubling

the input doubles the output

Operating State Independence

Starting

at y=0 is no different than starting at

y=20

Does not hold for nonlinear systems

SRSaunders - WSU - ChE 441

136

Symmetry
6

u1

u(t)

4
2

0
y(t)

-A

u2
0

0
-2
-4
-6

-2

-1

time

SRSaunders - WSU - ChE 441

137

Scaling
11
9
u2

2A
u(t)

7
u1

y(t)
0

5
3

1
-1
-2

time
SRSaunders - WSU - ChE 441

138

u(t)

Operating State Independence

u1

A
0
0

y(t)

9
8
7
6
5
4
3
2
1
0
-1
-2

time
SRSaunders - WSU - ChE 441

139

First-Order Systems (FOS)

Definition
Transfer

function with a first-order polynomial in

the denominator

Any

system governed by a first order differential

equation

This should look familiar

Liquid

Level Tank, Reactor in CSTR

SRSaunders - WSU - ChE 441

140

FOS - Properties

Gain
the

magnitude of the output change in

response to an input, denoted as K

Time constant
the

speed of the response after an input

change, denoted as

SRSaunders - WSU - ChE 441

141

FOS - Physics
Lets consider the tank
C

is an physical measure of resistance to flow

in the exit orifice
Find the gain and time constant for this
process

142

u(t)

FOS Step Response

u1

A
0
0

7
5
y(t)

3
1
-1

-1

time- WSU - ChE 441

SRSaunders

7
143

FOS Step Response

Characteristics
Final

Value

Gain

scales input

Initial

slope

Monotonic

output response

Each accomplishes
TIME
1
2

response
% of Final
Output

63.2%

SRSaunders - WSU - ChE 441

86.5%

of remaining
3

95%

98.2%

99.3%
144

FOS Practice
Using Matlab/Simulink, find the steady
state value and time to steady state for the
following FOS parameters in response to
Need

step input, transfer function and to

workspace blocks
Set simulation time to >100
Gain

10

10

40

40

Time
Constant

10

20

10

20

SS Value
Time to SS
SRSaunders - WSU - ChE 441

145

FOS Impulse Response

Vertical jump @ t=0

Followed by 1st order decay to SS
SRSaunders - WSU - ChE 441

146

FOS Rectangular Pulse

SRSaunders - WSU - ChE 441

147

FOS Pure Capacity

FOS

Pure Capacity: a0=0 => K*=b/a1

Called pure integrators

What is G(s)?
SRSaunders - WSU - ChE 441

148

u(t)

Pure Capacity Step Response

u1

A
0

y(t)

0
0SRSaunders - WSU - ChE 441

149

Pure Capacity Potential Safety

Risks

Lets look at a different tank with a fixed

outlet flow

Fin

h
SRSaunders - WSU - ChE 441

Fout
What happens if the
inlet valve is stepped
open just a little bit?
150

FOS Challenge Problem

cAo

Given these mixing

Draw the open loop

tanks

cA1
F

cA2
F

SRSaunders - WSU - ChE 441

block diagram for this

system
Calculate the effect of
a feed concentration
change (cAo) on the
exit concentration of
the second tank
Determine the order
Hint: Treat each tank
independently
151

Challenge Problem Solution

cA0

G1

cA1

G2

cA2

Tank 2

Tank 1

SRSaunders - WSU - ChE 441

152

Second-Order Systems (SOS)

Can result from two FOS in series
Higher

order systems possible by stringing

more FOS systems together

U1(s)

G1

Y1(s)
U2(s)

SRSaunders - WSU - ChE 441

G2

Y2(s)

153

SOS General Form

General second order system

Standard

form

SRSaunders - WSU - ChE 441

154

SOS Parameters
Three key parameters
Gain

K
Natural period
Damping coefficient

(Zeta)

Existence and nature of oscillation are

characterized by and
Poles?

SRSaunders - WSU - ChE 441

155

Lets look
response
K
2
2

15

10

y (t)

Which is :
Underdamped?
Overdamped?
Critically Damped?

at 3 test cases and their

to a unit step
10
10
10
40
42.25
13
25
13
3

K=10,t 2=40, 2tz=25

K=10,t 2=42.25, 2tz=13

K=10,t 2=40, 2tz=3

SRSaunders - WSU - ChE 441

10

20

30

40

50
time

60

70

80

90

156

100

SOS Dynamics - Damping Coefficient

Damping coefficient characterizes

qualitative process response
15

K=10,t2=40, 2tz=3

Case 1: 0 < < 1

are real or imaginary?
Displays
Has

10
y(t)

Poles

progression past the final value, 5

followed by a return to the
steady state
0

SRSaunders - WSU - ChE 441

20

40

time

60

80

157

100

SOS Dynamics - Damping Coefficient

Case 2: = 1

are real or
imaginary?
Fastest approach to
final value w/o
overshoot

10
8
6

y(t)

Poles

Case 3: > 1
Poles

are real or
imaginary?
Slower response
than case 2

2
0

Overdamped
Critically Damped
0

20

40

time

60

80

SRSaunders - WSU - ChE 441

158

100

SOS Dynamics
Damping Coefficient

Case 4: = 0

are real or
imaginary?
Oscillatory
response with no
damping

30
Oscillatory No Damping
Unstable

25

20

15

y(t)

Pole

10

10

20

30

40

50
time

60

70

80

Case 5: < 0

SRSaunders - WSU - ChE 441

159

90

100

FOS vs. SOS

Step response comparison
System

FOS

SOS

Final Value

AK

AK

Initial Value
Initial
SLOPE

Finite, nonzero

SOS

(w/ no zeros) always more

sluggish than FOS

SOS

SRSaunders - WSU - ChE 441

has an S shape response

TOS is even more sluggish
160

Underdamped SOS
Rise time
Time

to the first crossing of the final steady

state value

tr=12
A=1, K=10, 2=40, 2=3

SRSaunders - WSU - ChE 441

161

Underdamped SOS
Period
Time

between successive oscillation peaks

A=1, K=10, 2=40, 2=3

SRSaunders - WSU - ChE 441

162

Underdamped SOS
Decay Ratio
A

measure of the rate of oscillation decay

a1=4.64
a2=1.0

Overshoot
A=1, K=10, 2=40, 2=3
163

SRSaunders - WSU - ChE 441

Underdamped SOS
Settling Time time at which the output enters
(and remains within) a percentage of the final
value

Often 90%, 95% or 99% settling time

A=1, K=10, 2=40, 2=3

SRSaunders - WSU - ChE 441

164

FOS - Lead-lag Systems

System with a proper transfer function

Gain

Zero
Pole
Lead-to-lag

ratio

Lead arises from the zero, lag from the

pole.

SRSaunders - WSU - ChE 441

165

FOS Lead-Lag Systems

Partial Fraction Expansion

General Form

SRSaunders - WSU - ChE 441

166

Lead-Lag Step Response

Observations
For

very small t (t->0, i.e., use initial value

theorem) y(0)=

For

very large t (t->, i.e., use final value theorem)

y()=

Behavior

of y(t) is a big function of

SRSaunders - WSU - ChE 441

167

Lead-Lag Systems Effect of

Case 1: 0 < <
(0 < < 1)
Jump
@ moment of step
-> 0, becomes more
of a pure FOS and
lag dominates

y(t)

Discontinuous

A=1, K=5, =0.5, =1

0
0
SRSaunders
t - WSU - ChE 441

168

Lead-Lag Systems Effect of

Case 2: = ( = 1)
cancellation
Pure gain system
G(s)=K
Discontinuous jump @
moment of step

y(t)

Pole-zero

A=1, K=5, =0.5, =0.5

0
0
SRSaunders
t - WSU - ChE 441

169

Lead-Lag Systems Effect of

Case 3: > ( > 1)
Overshoot
dominates
Discontinuous jump @ moment of step

y(t)

Lead

0
0
SRSaunders
t - WSU - ChE 441

170

Lead-Lag Systems Effect of

Case 4: < 0 < ( < 0)
Inverse

response

Initial move away from the SS Value

y(t)

A=1, K=5, =-1, =0.5

0
0

SRSaunders - WSU - ChE 441

171

FOS In Parallel
G1

U(s)

G2

Y(s)

SOS with
Zeroes!
SRSaunders - WSU - ChE 441

172

SOS with Zeroes

2-pole, 1 zero system:
Output step response:

SRSaunders - WSU - ChE 441

173

SOS with Zeroes Step Response

A=1, K=10, =5, =10

>

SRSaunders - WSU - ChE 441

174

SOS Case Evaluations

Let 1 < 2, a > 0
Unless

otherwise noted

Case 1: a > 2

>

Case 2: a = 1 or a = 2

Yields

a FOS

SRSaunders - WSU - ChE 441

175

SOS Case Evaluations

Case 3: 0 < a < 2
Resembles

Case 4: a < 0
Always

displays

SRSaunders - WSU - ChE 441

176

Case Summary FOS a

Values

Key Observations

0 < a <

Jump at t=0 toward

y()
Pure gain system
(pole-zero
cancellation)

a =
a >

Overshoot

a < 0 <

Inverse response

Discontinuous jump @ t=0 for all cases

SRSaunders - WSU - ChE 441

177

Case Summary SOS a

Let 1 < 2, a > 0
Unless

otherwise noted

Values

Key Observations

0 < a < 2

Similar to FOS

a = 1 or 2
a > 2

FOS
(pole-zero cancellation)
Overshoot

a < 0

Inverse response

SRSaunders - WSU - ChE 441

178

Case Summary SOS

Values

Key Observations

<0

Unstable

=0
0<<1

Underdamped
oscillates forever
Overshoot and
underdamped

=1

Critically damped

>1

Overdamped
sluggish

SRSaunders - WSU - ChE 441

179

Inverse Respone
When a process output initially moves in a
direction opposite to its steady state value
followed by a return to steady state
Effect

of (at least) two opposing processes

are different timescales
Occurs when a < 0 in single-zero systems
y(t)

crosses the zero axis (in deviation

variables) in response to a step input

Where does this show up?

SRSaunders - WSU - ChE 441

180

A Two Timescale Exercise

Given the following block diagram and
transfer functions, calculate G(s)
G1

U(s)

G2

Y(s)

SRSaunders - WSU - ChE 441

181

FOS in Parallel
Two FOS in Parallel

Let:
|K1|>

|K2|
K1 and K2 be opposite signs
1 > 2 (G2 is faster than G1)

Consequences
Fast

process => Initial response

Slow Process => final response (due to higher
gain)

SRSaunders - WSU - ChE 441

182

Real Inverse Response

Drum Boiler
Used

for steam generation

Steam

Cold Water

Heat
Source

SRSaunders - WSU - ChE 441

183

Zeroes and Half Planes

Zeroes alters Partial Fraction Coefficients
=>

Altered

-1

Left Half Plane Zero

Speeds

system up -> Overshoot or jumps

Right Half Plane Zero

Leads

to inverse response

Multiple RHP Zeroes->

SRSaunders - WSU - ChE 441

184

Multiple RHP Zeroes

If # of RHP Zeroes is Odd
Classical

Inverse response -> initial move

away from steady state then towards the final
value

IF # of RHP Zeroes is even

System

displays inverse response, however

initial move is towards steady state
System response (y(t)) crosses y=0 a
number of time as there are RHP zeroes.

SRSaunders - WSU - ChE 441

185

Process Modeling
Models are rarely exactly known
Process model may not be available at all
Theoretical models may be very
complicated
Fundamental

model may require person-years

to develop

Less complicated models facilitate

controller design
What if a process model is unavailable?
Make

one from input-output data

SRSaunders - WSU - ChE 441

186

Process Identification
Technique for constructing process models by
relating input-output data
Black box modeling technique
Known
Manipulated
Input (u)

Process

Controlled
Output (y)
measured

Black Box
Given:

process data
Assume (or select): model structure
Calculate: model parameters
Evaluate: Performance (and remodel if
necessary)
SRSaunders - WSU - ChE 441

187

Process Identification
Things you should know
Properties
Explains

ONLY the input-output data

Tells you NOTHING about the actual process
physics
Extrapolation is fair (at best)

Model structure selection

Example:

You select FOTD

Assume that:
Linear process behavior covering the operation
range
Dynamics are (near) first order

SRSaunders - WSU - ChE 441

188

Theoretical Modeling vs. Process ID

Theoretical Modeling

Process ID

Few measurements

Lots of measurements
Only gain info about the
portion that the input affects
Treats the model like a black
box
Requires you to know
nothing about the actual
process
Sometimes the only option
Cannot easily make
nonlinear models

Only for unknown parameters

Provides info about the

internal state of the process
Promotes understanding of
the process
Requires accurate and
complete process
knowledge
Not good for complex
systems
Can easily produce linear
and nonlinear models
SRSaunders - WSU - ChE 441

189

Process ID Approach
Objective: given u(t) and y(t) data, find
Gp(s)

We have been doing, given u(t) and Gp(s) find y(t)

Step 1: Define the problem

Goals

for using the model?

How simple/complex does it need to be?
Relevant process aspects that need modeled?
Process well understood?
How do we measure model accuracy?
How much time do we have?
SRSaunders - WSU - ChE 441

190

Process ID Approach (2)

Step 2: Formulate the model
What type of (simple) model will be
used?
Study the process data
Force the process using input functions
Study process dynamics
Select

a model form

FOTD
Lead/Lag (1 pole, 1 zero)
SOTD
2-pole, 1 zero w/ delay
SRSaunders - WSU - ChE 441

191

Process ID Approach (3)

Step 3: Estimate model parameters
Time

domain methods: least squares,

statistical, graphical, etc.

Step 4: VALIDATE THE MODEL

Often

the forgotten step

How well does the model describe fresh data
What if a different forcing function is used?

Model is then ready

SRSaunders - WSU - ChE 441

192

Process Identification
The calculation of Gp(s) from u(t) and y(t)
data
Recall: the step of process ID
Define

the problem
Formulate the model
Estimate model parameters
Validate

What do we do for u(t)?

Use

the forcing functions

SRSaunders - WSU - ChE 441

193

Example Step Response ProID1

time
-1
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

u(t)
0
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5

y(t)
0
0
4.700124
8.847969
12.50843
15.73877
18.58954
21.10534
23.32552
25.28482
27.0139
28.53981
29.88642
31.07479
32.12353
33.04904
33.8658
34.58659
35.22268
35.78403
36.27942
36.7166
37.10241
37.44289
37.74335
38.00852
38.24252
38.44903
SRSaunders
38.63128

45
40
35
30

Input

25
Output
20
15
10
5
0
-5

15

Type of Model?

- WSU - ChE 441

35

55

75

95

Time (minutes)

194

Example 2 Graphical ProID2

time u(t)
y(t)
-5
0
0
0
0
10
3.16E-30 10
0
2.01E-05 10
0
1.21E-04 10
0
6.23E-04 10
0
0.003135 10
0
0.015694 10
0
0.078491 10
0
0.392477 10
0
1.962406 10
0
7.962406 10
0
13.96241 10
0
19.96241 10
0
25.96241 10 10.1792
31.96241 10 22.9713
37.96241 10 34.4251
43.96241 10 44.6805
49.96241 10
53.863
55.96241 10 62.0849
61.96241 10 69.4465
67.96241 10
76.038
73.96241 10 81.9399
79.96241 10 87.2243
85.96241 10 91.9559
91.96241 10 96.1924
97.96241 10 99.9857
103.9624 10 103.3821
109.9624 10 106.4232
115.9624 10 109.1462
121.9624 10 111.5843
127.9624 10 113.7672
133.9624 10 115.7219
SRSaunders
139.9624 10
117.472

140
120
100
80
Output

60
40
20
0
-5

45

95

145

195

245

295

Time (minutes)

Type of Model?

- WSU - ChE 441

195

Example 2 - Graphical
Model Structure (time domain, step
response)

Let ^ represent estimated quantity

Evaluate and simultaneously

SRSaunders - WSU - ChE 441

196

Example 2 Graphical Solution

ProID2

Find

140
120
100

80

Output

Plot and
evaluate for
and 12

60
40
20
0
-5

45

95

145

195

245

Time (minutes)

10
8
((( ))/ )

6
4
2
-5
0

95
f(x) =

195

Time

SRSaunders - WSU - ChE 441

295
197

295

Process Reaction Curve

Output

140
130
120
110
100
90
80
70
60
50
40
30
20
10
0
-10
-20
-30

-5

15

25

35

45

55

65

75

85

Time (minutes)
SRSaunders - WSU - ChE 441

198

95

Example Process Reaction Curve ProID3

35

Step of Magnitude 2
30
25
20
15
Output

10
5
0
-5
-10
0

10

20

30

40

50

60

70

80

90

Time (minutes)
SRSaunders - WSU - ChE 441

199

100

What About SOS?

Use the properties of SOS
Overshoot
Period

of Oscillation

Etc.

SRSaunders - WSU - ChE 441

200

What About Higher Order Systems?

First order with Time Delay (FOTD) models are

low order approximations
FOTD can be used to capture higher order
systems
Oberservations:
FOTD: K=24, =12,
=6
HOS: K1=2, K2=3,
K3=4, 1=5, 2=10,
3=3

SRSaunders - WSU - ChE 441

(+) reduces complexity

(+) facilitates controller
design
(-) not exact
(-) dynamic information
lost

201

Impulse Response ID
(Im)pulse input
Realistically: as short of a pulse as
possible yielding a sufficient output
Advantages
Operator/Manager

Friendly
System returns to original steady state
Should always yield a near linear response

Disadvantages
Small

perturbation -> small output -> noise

Less common than step testing

SRSaunders - WSU - ChE 441

202

Impulse Responses

Use Method of Moments

Utilizes

a statistical analysis
Tedious (but really good) for HOS
Easy for FOS

SRSaunders - WSU - ChE 441

203

FOS Impulse Response

Laplace Domain
Time Domain
jth Moment

SRSaunders - WSU - ChE 441

204

FOS Impulse Response

So what are K and ?

How do we calculate mj?

Evaluated

from data
Simpsons Rule?
Trapezoid Rule

SRSaunders - WSU - ChE 441

205

Example Impulse Response

time
-1
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9

0.35

output
0
0
0.3
0.32967
0.327864
0.326067
0.25
0.324281
0.322504
0.2
0.320737
0.318979
Output
0.317231
0.15
0.315493
0.313764
0.312045
0.1
0.310335
0.308635
0.306944
0.05
0.305262
0.303589
0
0.301926
0.300271
-1
0.298626
0.29699
0.295362
0.293744
0.292134
0.290534
0.288942
0.287358
0.285784
0.284218
0.282661
SRSaunders - WSU - ChE
0.281112

19

29

39

49

59

69

Time (minutes)

441

206

SOS Impulse Response

Use normalized process data

General form of SOS

Moments

SRSaunders - WSU - ChE 441

207