E N G I N E E R I N G I S T H E A R T O F M O D E L L I N G M AT E R I A L S W E D O N O T W H O L LY

U N D E R S TA N D , I N T O S H A P E S W E C A N N O T P R E C I S E LY A N A LY S E S O A S T O

W I T H S TA N D F O R C E S W E C A N N O T P R O P E R LY A S S E S S , I N S U C H A W AY T H AT

THE PUBLIC HAS NO REASON TO SUSPECT THE EXTENT OF OUR IGNO-

RANCE.

– D R . A . R . DY K E S , B R I T I S H I N S T I T U T I O N O F S T R U C T U R A L E N G I N E E R S

A S S U M E T H E C O W I S A S P H E R E T O M A K E T H E M AT H E A S I E R .

– UNKNOWN

W E D E M A N D R I G I D LY D E F I N E D A R E A S O F D O U B T A N D U N C E R TA I N T Y !

– DOUGLAS ADAMS, THE HITCHHIKER’S GUIDE TO THE GALAXY

D R . J U L I E VA L E

ENGG*2400
ENGINEERING
S Y S T E M S A N A LY S I S

WINTER 2019
Copyright © 2019 Dr. Julie Vale

Much thanks to Dr. Bob Dony for his assistance in providing content and in editing the document and to
Dr. Cam Farrow for his input into content organization, assignment problems, and course focus.
Contents

1 Systems: an Introduction 9
1.1 What is a System? . . . . . . . . . . . . . . . . . . . . . . . 10 Text: Ch. 1
1.1.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . 11
1.1.2 Modelling requires assumptions! . . . . . . . . . . 12
1.1.3 Through, Across, and Elements . . . . . . . . . . 13
1.2 Assignment . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2 First Order Systems 17

2.1 Fluid Systems . . . . . . . . . . . . . . . . . . . . . . . . . 17 Text: Chapter 12
2.1.1 Hydraulic Elements - Pipes/Hydraulic Resistance 19
2.1.2 Hydraulic Elements - Storage Tanks/Capacitance 21
2.1.3 The big picture - hydraulic elements . . . . . . . . 22
2.1.4 Summary of first order hydraulic systems . . . . 23
2.1.5 Assignment . . . . . . . . . . . . . . . . . . . . . . 23
2.2 Responses . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Text: pages 200- 215
2.2.1 Assignment . . . . . . . . . . . . . . . . . . . . . . 28
2.3 Thermal systems . . . . . . . . . . . . . . . . . . . . . . . 29 Text: Chapter 11
2.3.1 Thermal Elements - Resistance . . . . . . . . . . . 30
2.3.2 Thermal Elements - Capacitance . . . . . . . . . . 31
2.3.3 The big picture: Thermal elements . . . . . . . . . 32
2.3.4 Summary of first order thermal systems . . . . . 32
2.4 Assignment . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.5 Flow Diagrams . . . . . . . . . . . . . . . . . . . . . . . . 35
2.5.1 Assignment . . . . . . . . . . . . . . . . . . . . . . 38
2.6 Inputs and outputs . . . . . . . . . . . . . . . . . . . . . . 39
2.6.1 Breaking up large problems - identifying subsys-
tems . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.6.2 Modelling the milk tank . . . . . . . . . . . . . . . 41
2.6.3 Assignment: . . . . . . . . . . . . . . . . . . . . . . 43
2.7 System responses - Part 2 . . . . . . . . . . . . . . . . . . 45 Text: pages 200-201
2.7.1 No inputs - the Free Response . . . . . . . . . . . 47
2.7.2 Forced Response - introduction . . . . . . . . . . . 48
2.7.3 Impulse response of a first order system . . . . . 49
6

2.7.4 Step response of a first order system . . . . . . . . 51

2.7.5 Complete response . . . . . . . . . . . . . . . . . . 53
2.7.6 Assignment . . . . . . . . . . . . . . . . . . . . . . 55
2.8 Electrical Systems and the Big Picture . . . . . . . . . . . 57 Text: Ch. 6
2.8.1 Impedance and the D operator . . . . . . . . . . . 60
2.8.2 The Big Picture Part 1 - First Order . . . . . . . . 61
2.8.3 H ( D ) . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.8.4 Summary of first order electrical systems . . . . . 65
2.8.5 Assignment: . . . . . . . . . . . . . . . . . . . . . . 66

3 Interesting things we can do with H ( D ) 69

3.1 Nodal analysis . . . . . . . . . . . . . . . . . . . . . . . . . 69 Text: Section 6.4
3.1.1 Assignment: . . . . . . . . . . . . . . . . . . . . . . 74
3.2 Block Diagram Manipulation . . . . . . . . . . . . . . . . 75 Text: Chapter 13
3.2.1 Simplifying block diagrams . . . . . . . . . . . . . 75
3.2.2 Assignment: . . . . . . . . . . . . . . . . . . . . . . 80
3.3 Frequency response . . . . . . . . . . . . . . . . . . . . . . 81 Text: Section 8.4
3.3.1 Assignment: . . . . . . . . . . . . . . . . . . . . . . 87

4 Second order systems 91

4.1 Mechanical systems and Free Body Diagrams . . . . . . 91 Text: Ch 2
4.1.1 Mechanical Elements . . . . . . . . . . . . . . . . . 92
4.1.2 Free Body Diagrams and D’Alembert’s law . . . . 94
4.1.3 FBD for systems with more than one mass . . . . 96
4.1.4 Assignment: . . . . . . . . . . . . . . . . . . . . . . 97
4.2 Energy exchange, resonance, and inertiance . . . . . . . 98 Text: Ch 2
4.2.1 Resonance . . . . . . . . . . . . . . . . . . . . . . . 99
4.2.2 Inertiance . . . . . . . . . . . . . . . . . . . . . . . 100
4.2.3 Assignment: . . . . . . . . . . . . . . . . . . . . . . 104
4.3 State Variables . . . . . . . . . . . . . . . . . . . . . . . . 105 Text: Sections 3.1, 6.6
4.3.1 State-space and Matlab . . . . . . . . . . . . . . . 109
4.3.2 Assignment: . . . . . . . . . . . . . . . . . . . . . . 110
4.4 Second order step and impulse responses . . . . . . . . 111 Text: Subsection starting on p258
4.4.1 Characteristic Polynomial . . . . . . . . . . . . . . 112
4.4.2 Real roots: over and critically damped . . . . . . 113
4.4.3 Complex roots: under and un-damped . . . . . . 114
4.4.4 Summary of damping . . . . . . . . . . . . . . . . 119
4.4.5 Assignment: . . . . . . . . . . . . . . . . . . . . . . 120
4.5 Big Picture - Second order . . . . . . . . . . . . . . . . . 122
4.5.1 Assignment: . . . . . . . . . . . . . . . . . . . . . . 125

5 Laplace Transform 127

5.1 Definitions and Transfer Functions . . . . . . . . . . . . 127 Text: Sections 7.2 and 7.3
5.1.1 Transfer functions and the free and forced responses129
5.1.2 Relationship to state space (time permitting) . . . 132
 7

5.1.3 Assignment: . . . . . . . . . . . . . . . . . . . . . . 134

5.2 Partial Fraction Expansion . . . . . . . . . . . . . . . . . 135 Text: Section 7.6
5.2.1 Assignment: . . . . . . . . . . . . . . . . . . . . . . 139
5.3 ‘Black box’ modelling . . . . . . . . . . . . . . . . . . . . . 140
5.3.1 Assignment: . . . . . . . . . . . . . . . . . . . . . . 142
5.4 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 Text: pg 200, 253-257
5.4.1 Bounded Input Bounded Output . . . . . . . . . . 146
5.4.2 So why does it matter? . . . . . . . . . . . . . . . 147
5.4.3 Assignment: . . . . . . . . . . . . . . . . . . . . . . 148
5.5 Final Value Theorem and Static Gain . . . . . . . . . . . 150 Text: pg 233-235
5.5.1 Static Gain . . . . . . . . . . . . . . . . . . . . . . . 152
5.5.2 Assignment: . . . . . . . . . . . . . . . . . . . . . . 154
1
Systems: an Introduction

Welcome to ENGG*2400: Engineering Systems Analysis.

These notes are designed to provide you with a basis for taking your These notes are not a replacement for
own notes in class: some sections are left blank for you to fill in, coming to class! If you skip lectures
under the assumption that you can sim-
and I will often discuss details that are not in these notes. I strongly ply fill in the blanks from a colleague’s
suggest that you print out these notes and bring a hard copy to class notes, then there is a good chance that
you will miss important information.
with you. Sample problems for you to work through on your own That said, we are all adults: it is up to
time are at the end of each chapter; partial solutions will be posted you if you choose to skip lectures.
on courselink. A good secondary resource to the course text is the
book Modeling and Simulation of Dynamic Systems by Robert L.
Woods and Kent L. Lawrence. I have placed some copies on reserve
in the library. Text: When you see this notation, it
Note that this is the first iteration of these notes, so there will refers to chapters in the assigned course
text.
likely be some typos or errors. If you find one (or many), please Secondary Text: When you see this
contact me so that I can correct them. notation, it refers to chapters in the
book Modeling and Simulation of
The style of these notes is taken from Edward Tufte1 I hope that Dynamic Systems.
you find these notes as clear and readable as I find his work. The 1
I used the LaTex style found at
main text provides the core concepts, while illustrative figures, com- http://code.google.com/p/tufte-latex/.
mentary, etc. will appear in the sidenotes. I warmly welcome any
feedback on how I can improve these notes, including comments on
style, structure, organization, and content.

The following is a list of basic mathematics that I expect

you to know and be able to do quickly and without needing to think
about it. Some of these may seem obvious, but each and every item
on this list is something that more than one student has asked me
about in past terms. If you find that it takes a lot of time
to do these problems or that you get
• Complex numbers (as seen in ENGG*1500). totally stuck on these ideas then you
will spend too much time worrying
• Basic algebra; for example, about the math and not enough time
understanding the systems. It is in your
1. best interest to ensure that your math
V V skills are strong.
V = IR ⇔ =I⇔R= ,
R I
10 engg*2400 engineering systems analysis

2.
x ( a + b) = xa + xb, and

3. solve the following for y:

x
+a=q
y+z
x + (−q + a)z
⇔ = y.
q−a

• Ability to plot and interpret functions, specifically exponential and

sinusoid (e.g. f (t) = e−t )

• Basic trig including sohcahtoa and manipulating right angle trian-

gles in the four quadrants of a 2-dimensional plane.

• The ability to manipulate fractions quickly and efficiently; e.g.,

x a xb + ya
+ = , and
y b yb
a
b ayz
x w = .
y + z xbz + wyb

Beyond the examples above, I expect you to be able to work with

variables, numbers, or some combination of both. It is also crucial
that you use units in your solutions (where appropriate): this may
occasionally require unit conversion.

1.1 What is a System? Text: Ch. 1

By the end of this section, you will be able to Secondary Text: Section 1.1

• Explain, in your own words, what a system, boundary, parameter,

input, and output are.

• Discuss some common assumptions and why/when it is impor-

tant to use them.

• In your own words, explain what an element is and what across

and through variables are.

This course provides the key underlying concepts of Sys-

tems Theory, which is a broad area of study focusing on
the interactions of many different types of devices, pro-
cesses, or mechanisms. The field came to prominence in the mid-
nineteen hundreds during the push to develop rockets and space
flight and was integral to the success of those programs. Since then,
 systems: an introduction 11

systems theory has broadened to be used in any engineering project

that requires interaction between the more traditional ’silos’; indeed, For example, the design of airport
many large engineering design firms have systems departments radar requires civil works, mechanical
devices, electrical and signal processing
whose goal is to integrate the disparate parts of the design into one design, and software integration
coherent whole.
In this course, we’ll only scratch the surface of this field. We’ll
focus on mechanical, electrical, thermal, and fluid systems. We’ll
model these systems with the goal of understanding how all four are
actually very similar (even though they seem like they’re not) and
about how to predict those system behaviours without needing to
solve the differential equations every single time. Key to the field is the idea that sys-
Before launching into details of any particular system type, we’ll tems analysis is a way of approaching
complex problems by splitting them up
start by listing some key definitions and key ideas that will be used into simpler, interconnected, problems.
throughout the course. This course focuses on understanding
those simpler, unconnected problems;
connecting systems to one another is
1.1.1 Definitions touched on in the last quarter of the
course.

System: (From the Oxford English Dictionary)

• An organized or connected group of things.

• A collection of natural objects, features, or phenomena considered

as or forming a connected or complex whole.

• A group of natural objects moving in relation to one another under

the laws of nature.

Boundary: Chosen (usually by us!) to separate the system from its

‘environment’.
12 engg*2400 engineering systems analysis

Parameters: Variables (usually constants) that represent key prop-

erties of the system. We often don’t know these values. Typically,
we either approximate them based on known properties or they are
measured or determined through some kind of testing (which may be
destructive!). Either approach must account for variance/error.

Signals: Information or instructions delivered to, or measured from,

the system. Usually change with time.

Inputs: Signals that act on a system and cause it to react. User de-
fined inputs are applied to the system via an ‘actuator’, but not all
inputs are user defined.

Outputs: Signals that are a result of the actions of the input on the
system. Outputs of interest are often measured via a ‘sensor’.

1.1.2 Modelling requires assumptions!

A key aspect of modelling is knowing what kinds of as-

sumptions to make. For example, if we’re modelling the acceler-
ation of an entire car (and it isn’t a race car ;)), we may neglect the
mass of the fuel, but, if we’re modelling the flow of fuel from the gas
tank to the engine, then the mass of the fuel becomes crucial informa-
tion. Determining which assumptions to make depends heavily on
the system boundary and the signals of interest. Overly precise models are often useless
‘Obvious’ assumptions may include things like assuming uniform for a systems engineer: the added detail
obscures the crucial components that
gravity of 9.8m/s2 . Other examples? actually drive the dominant behaviours!

We often make ‘gross’ assumptions

We also tend to make ‘simplifications’ by neglecting things that that help with ‘back of the envelope’
type engineering. These are very useful
(we hope) have little significance; e.g., frictionless pulleys. Other for making ‘ballpark’ calculations
examples? and for predicting the rough order of
magnitude of the system’s behaviour.
They are particularly useful for ‘sanity
checking’ solutions.

In this course, we make the simplifying assumption that all sys- Text: See Ch. 9 for additional infor-
tems are Linear unless stated otherwise. You learned about linearity mation - you will not be required to
‘linearize’ models in this course.
in ENGG*1500.
 systems: an introduction 13

1.1.3 Through, Across, and Elements

Even when modelling something as simple as a single

tank system, we want to break that system down into
its fundamental parts, similar to how we can break chemical
molecules as a collection of elements. Indeed, even when we’re dis-
cussing fluids, or mechanics, or any other system model, these ele-
mentary components are called elements. Elements are chosen to be
those key components that are easy to model and that have both a
through and an across variable (usually signals) associated with them.

Through variables are those variables that are conserved in the

sense that what goes into the element is exactly the same as what
comes out of the element. These variables don’t change as they ’flow’
through the element. Examples include:

Across variables are those variables that are changed by the element
in some way, in the sense that what is measured on one side of the
element is different than the other. These variables yield a net differ-
ence when measured across the element. Think of these as variables
that express ‘effort’. Examples include:

These through and across variables can be used together with funda-
mental properties of the element to write an element law. We can express these element laws as
an impedance Z, yielding the relation-
ship Across = Through × Z.
14 engg*2400 engineering systems analysis

1.2 Assignment

This course relies heavily on mathematics; to that end, consider the following questions (modified
somewhat) from some old final exams. All of these questions can be answered using tools and techniques
that you learned in first year and in highschool. If you find any of these problems overly difficult or feel
like you have never seen these concepts before, I strongly suggest that you take the time to seek help from
your TAs or from the course instructors.

1. Perform the following calculations. In all cases, provide the answer in reduced form.
Rπ
a) 0 cos(t)dt
R
b) cos(t)dt
d(e at )
c) dt
Rt
d) 0 e−sτ dτ
e) 12 +x
− 624
17
f) ( x + 2y)(4x − y)
g) (s2 + 13s − 2)(s − 1)
h) Find the roots of the polynomial s2 − 3s + 12
√
i) Express 3 − 3 3j in polar form (you shouldn’t need a calculator)
j) Express (1 − j)4 in both polar and rectangular form (you shouldn’t need a calculator)
x +( a−w)z
2. Solve the following equation for a. Show your work. w− a =b
3. Consider the following system of equations relating an applied force f to a resulting position p:

ẋ1 (t) = 12γx1 (t) + γx2 (t) − 1.5 f (t)

1
ẋ2 (t) = − 4.3x1 (t) + f (t)
γ
p ( t ) = x1 ( t ) − x2 ( t ).

Convert those equations to their matrix equivalent by filling in the boxes below.
" # " #
ẋ1 (t) h i x (t)
1
h i
= + f (t)
ẋ2 (t) x2 ( t )
| {z } | {z }
=: ẋ (t) =:x (t)
h i h i
p(t) = x (t) + f ( t ).

4. Sketch and label the curves for the following functions, including what happens when the independent
variable is negative.

f 1 (t) = e2t , f 2 (t) = e−2t , f 3 (t) = 1 − e2t , f 4 (t) = 1 − e−2t ,

g1 ( x ) = sin( x ), g2 ( x ) = sin( x + π ), g3 ( x ) = cos( x ),

g4 ( x ) = 2 sin( x ), g5 ( x ) = sin(2x ), g6 ( x ) = 1 + sin( x ).
 systems: an introduction 15

5. Using graphical methods, sketch and label the integral of the curve shown below.

y1 ( t )

1 2 t

6. Using graphical methods, sketch and label the derivative of the curve shown below. On your sketch,
clearly indicate any locations where the derivative is ‘undefined’.

Current
8

-2

-4

-6

-8
0 2 4 6 8

7. You have tested a circuit in the lab and obtained the following voltage v and current i signals.
a) Using the equation power(t) = p(t) = v(t)i (t), sketch and label the power curve.
Rt
b) Using the equation energy(t) = w(t) = −∞ p(τ )dτ, sketch and label the energy curve.

Current Voltage
8
2
6

4
1.5
2

0 1

-2
0.5
-4

-6
0
-8
0 2 4 6 8 0 2 4 6 8
2
First Order Systems

This course focuses on dynamic systems, the simplest of

which are so-called ‘first order’ systems. Before we go too
far though, we should make sure that we understand what it means
to be ‘dynamic’ vs. ‘static’. A static system is one in which the output
at a specific time depends only on the input at the exact same time. We say that static systems have no
Ohm’s law and the ideal gas law are examples of static systems, memory, or in other words, the past
does not influence the present.
while any differential equation (where the derivative is a function of
time!) is dynamic.
If the system is static, then we don’t have to worry about solving
differential equations, which means that they become algebra prob-
lems that are (relatively) easy to solve. As soon as we need to worry
about energy storage (e.g., momentum), things become more compli-
cated. Let’s look at a simple fluid system to explore a simple example
of this energy storage phenomenon and see how it affects our system
model.

2.1 Fluid Systems Text: Chapter 12

By the end of this section, you will be able to: Secondary Text: Sections 5.3 and 5.4

• Identify the through and across variables and the fundamental

elements for hydraulic systems

• Calculate hydraulic Resistance and Capacitance

• Apply element laws to determine flow and pressure through and

across hydraulic elements
18 engg*2400 engineering systems analysis

Consider a very simple fluid system comprised of an open

tank and an outflow pipe: (i.e., a bathtub!) Note that modelling of fluid systems
is covered in much greater detail in
ENGG*2230. We’re looking at simple
systems here, and we’re going to be
making a lot of assumptions to help us
do so.

Signals of interest:

• The volume flow rate of the fluid q(t), usually in m3 /s. In the Flow is a through variable (more on this
figure, we have two arrows, each of which is a flow (we could add later!). Also, the text uses the variable ω
for flow instead of q.
more arrows to illustrate more flows, but we won’t do that right
now).

• The volume of the fluid V (t), usually in m3 . In this example, note

that dV (t)/dt = qin (t) − qout (t).

• The pressure P(t) at various locations in the bathtub, usually

in Pa or N/m2 . In the figure, where is the pressure the same as Pressure is an across variable (more on
atmosphere? Where is it different? this later!)

• The height of water in the tank h(t). Height is related to pressure Notice that both pressure and height
(we’ll see how very soon) and is often used to replace pressure can be measured relative to different
choices of ‘zero-reference’! For example,
in equations since we are typically more interested in how much we could measure pressure relative to
water is in the bathtub that we are in what the pressure is on the vacuum (0kPa), or we could measure
it relative to atmosphere (101kPa).
bottom of the tub. The latter is more common. What are
different zero-references for height that
might be relevant for this bathtub?
Across and through In Hydraulic systems, the through variable is
q, the volumetric flow rate (i.e., the mass flow of the fluid) and the
across variable is P, the pressure.

Okay, so now what? You haven’t completed the fluids course, so

where do we go from here? Let’s start by making some simplifying
assumptions:

Assumptions: Assume the fluid is incompressible and that all flow is

laminar.

That’s nice.... but what does that even mean??

 first order systems 19

Incompressible means that the fluid does not change volume when
put under pressure. Water is incompressible (the volume does change
with pressure, but that change is negligible unless the pressure is
huge), but air is not.

Laminar flow is ‘nice’ in the sense that the water particles flow in
clean, parallel streams. When the water in your tap is running clear,
it is laminar. The other type of flow is called ‘turbulent’; here, parti-
cles dance and swirl around a lot. Modelling turbulent flow is very
difficult and requires complex, highly nonlinear equations, so we’re
going to avoid it. You’ll learn more in your Fluids course. We can identify whether or not flow is
laminar by using something called the
Reynolds number, which is a unit-less
There are two elements of interest in our bathtub example:1 the number that is based on a combination
hydraulic resistance (in the pipe), and the open tank. of the fluid’s properties and the pipe’s
properties. We tend to get laminar flow
when the fluid is slow or when the pipe
2.1.1 Hydraulic Elements - Pipes/Hydraulic Resistance has a large diameter.
1
There’s a third hydraulic element, but
we’ll learn more about that one after
the midterm.
Let’s look at the resistance first, but, before getting into the
gory details, let’s take a moment to figure out (at a high level) what’s
going on. Consider a simple, cylindrical pipe:

The pressure difference from one end of the pipe to the other is like a
‘push’2 . If the pressure on one end of the pipe is the same as on the 2
Remember that Pressure is Force per
other end (i.e., P+ = P− ), the fluid won’t move (flow will be zero) unit area!

because there is nothing ‘pushing’ the fluid through that pipe. Now,
what if P+ is larger than P− ? Which direction will the fluid flow?
Okay, so now we have a relationship between the direction of flow
and the pressure drop. Next, we need to figure out the details of that
relationship. To do so, we need to acknowledge that the fluid will
experience friction on the sidewalls of the pipe as it flows through
that pipe, and that friction will resist the flow of the fluid. More
resistance means slower flow OR a higher pressure drop (i.e., you
need to push harder to get the fluid to flow at the same speed!). It
turns out that this resistance R is a function of the dimensions of the
pipe and of the fluid properties:

• The absolute (or ‘dynamic’) viscosity of the fluid µ, usually in Pa s

• The diameter of the pipe d, usually in m

• The length of the pipe `, usually in m.

20 engg*2400 engineering systems analysis

• Cross sectional area A of the pipe, usually in m2 .

Putting these together yields

32µ`
R= .
Ad2
For the case where the pipe is a cylin-
Great, so how do we use this resistance? Well, jumping through a der, can you simplify this expression
to remove A? (Hint: yes, you can! Try
few hoops and some assumptions about linearity yields the following it...) What happens if the pipe isn’t a
element law: cylinder? You’ll learn how to handle
this in your fluids course...
1
q(t) = ( P+ (t) − P− (t)) ⇔ ∆P(t) = Rq(t).
R| {z }
∆P(t)

Example: The city of Guelph is supplied with approximately 47 km3 /day

of water. They pipe in water from local aquifers at a rate of 4.38 m3 /s.
The longest pipe is approximately 2km long and 1m in diameter.
From tables, you find that the kinematic viscosity of water is 10−6
m2 /s and the dynamic viscosity of water is 10−3 Pa s. What is the
overall pressure drop from one end of this long pipe to the other? As you work through this problem,
Start by calculating the pipe resistance R. ask yourself what assumptions you’re
making and whether or not the answers
you’re getting seem to make sense.

Now use that to find the pressure.

Does this answer make sense to you? Remember, if the weight3 of an 3

We typically use the word ‘weight’
apple on the surface of the Earth is approximately 1 Newton, then incorrectly (e.g., when we talk about
how much we ‘weigh’ we should be
the pressure exerted by that apple on the palm of your hand is a talking about our ‘mass’). Weight is
little more than 100Pa... If the answer to the fluid problem does make the force of gravity acting on mass, so
weight is a force... if you don’t believe
sense, great. If not, where did we go wrong? me, Google it ;)
 first order systems 21

2.1.2 Hydraulic Elements - Storage Tanks/Capacitance

Now it’s time to take a closer look at the tank. The most
important thing to keep in mind is that a tank is a storage device,
but what does it store? To answer this question, ask yourself ‘what is
accumulating in the tank’?

We’ll need the following variables in this discussion:

• The density of the fluid ρ, usually in kg/m3 .

• Cross sectional area A of the tank, usually in m2 .

• Volume of fluid in the tank V, usually in m3 .

• Mass of fluid in the tank m, usually in kg.

• Gravitational constant g = 9.8m/s2 .

And it would be useful to have a labelled figure for reference:

We’ll only consider open4 , vertical-walled5 tanks in this course, 4

This lets us assume the surrounding
so the pressure at the surface of the water is always the atmospheric air has no effect on the behaviour of the
system.
pressure. So what’s the pressure at the bottom of the tank? What 5
This means the volume has a simple
does it depend on, and how does it change? And how does that relationship to height: V = Ah, with A
pressure relate to the flow in and the flow out from the tank? a constant instead of A depending on h.

It turns out that some of the above questions are easier to answer
than they first appear. Let’s start by relating pressure to height by
using the force of the water acting on the bottom of the tank: F = mg;
except we really want the pressure, which is force per unit area, so
let’s divide: F/A = mg/A = Pb − Patm = ∆P. In fluids, we prefer to Why are we using ∆P instead of just
deal with fluid properties and object properties, so let’s remove m by the pressure at the bottom of the tank?
Because we’re only using the force due
to the water, not the force due to the
water and the entire atmosphere above
it! Next question: Why is it Pb − Patm
instead of the other way around? Well...
does adding fluid make the pressure
higher or lower? Based on that, which
choice ensures we get the right sign?
22 engg*2400 engineering systems analysis

using the fluid’s density ρ = m/V, yielding

mg ρVg
∆P = = = ρgh.
A A
Great, now we can relate pressure to height. Why does this mat-
ter? Because height is much easier to visualize, especially when we’re
trying to find a relationship between pressure and flow (which we
need to do to determine our element law), and because height is of-
ten an output signal of interest. Consider the total flow through the
tank q6 . If q > 0, then the tank is filling; if q < 0, then the tank is 6
By ‘through’ the tank, we mean the
emptying; and if q = 0, then the tank volume (or height) is staying sum of all the flow into the tank minus
the sum of all the flow out of the tank.
the same (which we call steady). Furthermore, since q is defined as
the change in volumetric flow per unit time q = V̇, we can use our
derived relationships to find the element law

A
q(t) = V̇ (t) = A ḣ(t) = ∆ Ṗ(t)
ρg
|{z}
=:C

and the associated Hydraulic Capacitance C.

2.1.3 The big picture - hydraulic elements

Across Through Resistive element Capacitive element Capacitor stores?

Hydraulic
 first order systems 23

2.1.4 Summary of first order hydraulic systems

Assumptions:

• The fluid is incompressible and all flow is laminar.

• All tanks have vertical walls.

• All tanks are open to the air.

Variables: The through variable is the volumetric flow rate q and the
across variable is the pressure P.

Summary of parameters of interest: These values are typically ob-

tained from tables of properties or by physically measuring the di-
mensions of the system components.

• The density of the fluid ρ, usually in kg/m3 .

• The absolute (or ‘dynamic’) viscosity of the fluid µ, usually in Pa s.

• The kinematic viscosity of the fluid ν, usually in m2 /s.

• The diameter of the pipe d, usually in m.

• The length of the pipe `, usually in m.

• Cross sectional area A, usually in m2 . May be of a pipe or of a

tank, depending on context.

Element Laws
1 128µ`
• Pipe resistance: q = R ( P+ − P− ), R = πd4
A
• Open Tank: q = C ( Ṗ+ − Ṗ− ), C = ρg

• Fluid inertia: TBD

Sources: Pumps (TBD) We haven’t quite done this yet, but we

will soon. I’ve included it here for your
• Independent: Provide known flow q or known pressure difference reference for later
∆P.

• Dependent: Flow depends on some nominal value q̄ and one or

more pressures elsewhere in the system

2.1.5 Assignment
None - see next lecture
24 engg*2400 engineering systems analysis

2.2 Responses Text: pages 200- 215

By the end of this section, you will be able to:

• Find a differential equation that models a simple, first order hy-

draulic system

• Explain, in your own words, what a response is

• Explain, in your own words, what a forcing function is

• Explain, in your own words, what an initial condition is

• Explain, in your own words, the difference between the steady

state and transient part of a response Secondary Text: pg 11

In today’s lecture we’re going to try to understand how

the large tank from last day responds to various outside
effects. To that end, our goal for the first part of this lecture is to
find a differential equation that models the relationship between the
pressure at the bottom of the tank and qin , assuming that the tank is
draining through a rough pipe. Before you begin, make a plan! What
do we know? How can we connect the dots between all the different
things that we know to come to a final answer that relates h to q? For
the purposes of this example, assume that you’ve already calculated
the capacitance C and the resistance R and that the flow is laminar.
 first order systems 25

Now we have a differential equation, so what? Well, one

of the main goals of this course is to be able to predict
how a system will behave to certain inputs without need-
ing to do a lot of math. Unfortunately, in order to do this, we
actually need to do the math a few times so that we can get a han-
dle on some of the key patterns and behaviours that our systems
will share. To do so, we need to discuss and define some important
terms that will come up quite often in this course and in many other
courses after this one.

Forcing function: We are almost always interested in how a system

will react to certain types of inputs. If we know what the equation is
for that input (e.g., maybe the input is an applied force and that force
follows the curve f (t) = e−2t ) then we can call f (t) a forcing function.

Initial Condition: As engineers, we usually don’t know what our

system was doing before we started looking at it. It turns out that
(for linear time invariant systems), we don’t really need to worry
about the past history of the system behaviour because it can be
entirely encapsulated into a single measurement: the initial conditions.
You’ll be seeing a lot of this in your calculus course in the context of
‘initial value problems’ or IVPs.

Response: A system’s response is the output of that system due to

a particular input (or forcing function) combined with the initial
conditions.

Example: Let’s consider a overly simplified problem to illustrate

these ideas. It’s February. You arrive home after a late night study
period. When you left in the morning, the room was a comfortable
21o C. When you get back home, the temperature in the room is 15o C.
You turn on the heater, which pumps out heat at a rate of 100MJ/hr.
Because you’re in ENGG2400, you measure the temperature every
minute and plot the resulting curve, which turns out to satisfy an
equation that looks (roughly) like a − be−t/τ (with a, b, and τ param-
eters chosen to fit the curve). What is the initial condition, the forcing
function, and the response?
26 engg*2400 engineering systems analysis

Steady state vs Transient: These are paired ideas! The transients are
the pieces of a function (or signal) f (t) that decay to zero. A sys-
tem response tends to have transient components corresponding
to changes in the forcing function changes (e.g., the forcing func-
tion changes from a sinusoid to a constant). The steady state portion
of a function f (t) is the part that is still sticking around after the It is possible for a signal to have no
steady state component (e.g., if the
transients are (more or less) gone. The steady state must be either
signal goes to infinity, such as et ).
constant or periodic.

Example: y(t) = e−2t + 6 sin(t), t ≥ 0.

transient steady state

6

6sin(t)
-2

-4
e(-2t)+6sin(t)

-6
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Great, now back to the bathtub example. Remember that we found

the following differential equation: RP(t) + C Ṗ(t) = qin (t). It turns
out that, if the inflow is a constant, then the solution to that differ-
ential equation will satisfy the same curve as the room temperature
problem:7 7
You’ll be able to derive this yourself by
the end of this course!
q  q 
a − be−t/τ = in + p(0− ) − in e−( R/C)t .
R R
Assuming that all of the variables are positive numbers, sketch and
label the curve and identify the transient and steady state compo-
nents. Sneak preview: The portion of the
response due to the input qin is called
the forced part, and the part due to the
initial condition is called the free part!
 first order systems 27

So what does this tell us about our tank’s behaviour?

What aspect of the tank determines the steady state part of the re-
sponse?

What about the transient part of the response? (Think about the
‘speed’ of that response)

Why does this make sense?

What do you think will happen if we turn the water off?

28 engg*2400 engineering systems analysis

2.2.1 Assignment
1. Show that, for a cylinder,
128µ`
R= .
πd4
2. From the text, consider the problem outlined in 12.5 (pg 413). Consider the flow w0 to be the output.
Find the differential equation modelling this system. Note that Pa is atmospheric pressure and the incre-
mental pressures are defined as p̂i := pi − p a . Note also that this question introduces the idea of a pump,
which you can read about on page 403. For now, just use the equation of the pump that you’ve been
provided and know that the flow rate into the pump is equal to the flow rate out of the pipe.
3. In the following output signals, identify the transient and the steady state components. Sketch the
curves and (where appropriate) indicate where the transient component is (approximately) ‘gone’.
a) y(t) = e−2t + 6, t ≥ 0.
b) y(t) = 3(e−2t + 6), t ≥ 0.
c) y(t) = e2t + 6, t ≥ 0.
d) y(t) = 1 − e−2t , t ≥ 0.
 first order systems 29

2.3 Thermal systems Text: Chapter 11

By the end of this section, you will be able to: Secondary Text: Section 6.2

• Identify the through and across variables and the fundamental

elements for thermal systems

• Calculate thermal Resistance and Capacitance

• Identify common sources in thermal systems and their associated

properties

• Apply element laws to determine heat flow and temperature

change through and across thermal elements

Consider a very simple thermal system comprised of an enclosed,

insulated space: (e.g., a hot water tank!) Note that modelling of thermal systems
is covered in much greater detail in
ENGG*3260.

We can ask lots of questions about this tank. E.g., one common ques-
tion is how can we determine how long it will take the water in the
tank to warm up or cool down to a pre-specified temperature? To
answer this question, we’ll need a model.

Assumptions: Assume that the temperature in any vessel is uniform.8 8

This isn’t true... e.g., of course it will
We have to be very careful when choosing our temperatures. We can be warmer closer to the heater, but
modelling the temperature gradient
use absolute temperature (in which case we need to use Kelvin), or inside the tank is well beyond the scope
we can take a temperature relative to some known value (e.g., the of this course. You may also hear the
phrase ‘assume the fluid is perfectly
ambient temperature). For example, if we want to use Celsius instead mixed’.
of Kelvin, what is our measurement relative to (in Kelvin)? What if
we want to measure how much warmer or cooler we are than room
temperature?
30 engg*2400 engineering systems analysis

2.3.1 Thermal Elements - Resistance

There are a number of ways that heat can move, but in the
course, we’ll care about two: conduction and forced convection. Both Other methods of heat transfer include
of these heat transfer methods can be modelled as a thermal resis- free convection, black body radiation,
etc.
tance. For conduction, we’re essentially interested in heat transfer
through an object (like a wall): when we apply heat to one side of
this object, what is the heat flow to the other side? For convection,
we’re looking at the effect of fluid flow against a surface to model the
change in heat on that surface. Text: Convection is not covered in this
In both cases, just like with hydraulic systems, we can find a ther- text
Secondary Text: Section 6.2.2
mal resistance R that satisfies the following element law: Derivations of thermal resistance are
beyond the scope of this course. You’ll
∆T (t) = ( T+ (t) − T− (t)) = Rq(t) learn more in thermo!

So, how do we find R? The assumption here is that T+ is the

Let’s start with conduction. The actual value of R depends on the hot side and that heat flows from hot to
cold.
geometry of the object and the material properties of that object. For
example, if it is a flat plate, then, using

• The thickness of the object d, usually in m

• Surface area A of the object, usually in m2 .

• The thermal conductivity of the material α, usually in W/(m K ).

we get These equations for R are derived from

d Fourier’s equation and assume that the
R= . temperature gradient in the material is
Aα linear. This assumption holds when the
On the other hand, if we’re interested in the heat transfer through the material is relatively thin and when the
walls of a pipe (for example) then, with temperature change ∆T is not large.

• The inner radius is ri , usually in m

• The outer radius is ro , usually in m

• The length of the pipe is `, usually in m

• The thermal conductivity of the material α, usually in W/(m K ).

we get
ln(ro /ri )
R= .
2πα`
Forced convection yields a similar result for R. We’ll need the
following parameters:

• Surface area A of the object, usually in m2 .

• The convection coefficient h, usually in W/(m2 K ).

 first order systems 31

In which case we get

1
R= .
Ah
This resistance is usually quite difficult
In all cases, R has units K/W (or C/W... they’re equivalent!). to calculate and is determined through
experimentation (and therefore looked
up in tables).
Example: A simple heat exchange element can be modelled as a
resistive element with thermal resistance R. Sketch the element, la-
belling all important variables, then write down equations that model
the system. (Hint: Remember to think about conservation too!)

2.3.2 Thermal Elements - Capacitance

Why does the cheese on a pizza take longer to cool down

than the crust? Because of a property called thermal capacitance;
i.e., in some sense, the cheese is better at ‘storing’ temperature than
the crust is. Key to understanding how this works is remembering
that any object (or quantity of fluid) has a certain amount of thermal
energy that relates to both the material and to temperature. We’ll
need a few parameters:

• The mass of the object m, usually in kg

• The specific heat of that object C p , usually measured in Joules/(kg

Kelvin) - J/(kg K)

Great. Now we can say the following: Thermal Energy is given by

E(t) = mC p T, note that T is in Kelvin here!

We can use this concept to derive a relationship between the tem-

perature of an object and the heat flow to (or from) that object. By
definition, heat flow is the change in thermal energy over time, so we
have
dE(t) d mC p T
q(t) = = .
dt dt
32 engg*2400 engineering systems analysis

There’s a time derivative here, so now we have to ask an important

question: for our pizza cheese example, which of the three variables
change with time and which are constant?

What if we don’t want to use Kelvin?

Let’s see what happens: let Tc =
Okay, so in that case, we have T − 273.15 (i.e., Tc is the temperature in
Celsius). Then we have
d mC p T (t) dT (t) Tc = T − 273.15
q(t) = = mC p ,
dt dt ⇔ T = Tc + 273.15

since temperature is our across variable and flow is our through, we ⇔ Ṫ = Ṫc + 0 = Ṫc

can write this as an element law: Therefore, we have

q(t) = mC p Ṫc (t).
q(t) = mC p Ṫ (t).
|{z} So we can use Kelvin or Celsius (or a
=C different, relative measure of tempera-
ture) in this element law... nice!

2.3.3 The big picture: Thermal elements

Across Through Resistive element Capacitive element Capacitor stores?

Hydraulic

Thermal

2.3.4 Summary of first order thermal systems

Assumptions:

• Linear temperature gradients in materials

• Perfectly stirred tanks

 first order systems 33

Variables: The through variable is the heat flow q and the across
variable is the temperature T.

Summary of parameters of interest: These values are typically ob-

tained from tables of properties or by physically measuring the di-
mensions of the system components.

• The thickness of the object d, usually in m

• Surface area A of the object, usually in m2 .

• The inner radius is ri , usually in m

• The outer radius is ro , usually in m

• The length `, usually in m

• The thermal conductivity α, usually in W/(m K ).

• The convection coefficient h, usually in W/(m2 K ).

• The mass m, usually in kg

• The specific heat C p , usually in J/(kg K )

Element Laws

• Thermal resistance: Rq = ( T+ − T− )
d
– Conduction flat plate R = αA
(ro /ri )
– Conduction radial pipe R = ln2πα `
1
– Forced convection R = hA

• Thermal capacitance: q = C ( Ṫ+ − Ṫ− ), C = mC p

• There is no Thermal inertia!

Sources:

• Ambient temperature: provides known Temperature Ta .

• Mass flow: provides known heat flow satisfying q = ṁC p T =

wC p T.

• Heating elements: provides known heat flow q

34 engg*2400 engineering systems analysis

2.4 Assignment

None
 first order systems 35

2.5 Flow Diagrams

By the end of this section, you will be able to:

• Construct a flow diagram for a system

• Find a differential equation that models a first order thermal sys-

tem

We’re going to use a single example (a hot water tank) throughout

this lecture to illustrate some important concepts.

Example: You’ve just finished a shower and used up all of the hot
water. Your roommate needs to know how long to wait until they can
shower.
We’ll use ENGG*2400 techniques to solve this problem. Let’s start
by finding a relationship between the heater element and the tem-
perature of the water in the tank, using variables R and C in place of
numeric values.
But wait.... we’re still missing some important information: what is
the impact of the cold water inflow? For simplicity, let’s assume that,
as we remove water from the tank (for our shower), the water gets
replaced by cold water from the city’s system, maintaining a constant
volume of water in the tank. If we go back to our energy equation
with this assumption, then we have

d mC p T dm There are a few important things to

q(t) = = Cp T .
dt dt point out about this equation: 1. T is
the absolute temperature of the fluid
Just like in our fluids example, let’s replace mass flow ṁ with the that is flowing, not the temperature
volumetric flow rate of the fluid in the tank. 2. Here q
represents heat flow, not volumetric
flow rate... when discussing thermo,
q(t) = ṁ(t)C p T = ρC p Tw(t). let’s use w for the volumetric flow rate.
Recall ρV = m and V̇ = w, so ṁ =
What does this mean? ρV̇ = ρw
Note that an implicit assumption here is
that the mass flow rate ṁ is constant, so
V̇ is also constant, as is w.

What about the ambient temperature?

What other assumptions should we make?

36 engg*2400 engineering systems analysis

Let’s worry about finding the actual values of the R and C terms To make a flow diagram, identify
each element and label the across and
later. For now, let’s proceed with finding a parametric model by
through variables for that element.
constructing a flow diagram. Then, look for branches that have the
same across variable and connect them
up into nodes. Indicate the direction of
flow with an arrow, and you’re done.

Now write the element laws and the conservation laws (based on
the flow diagram), and simplify the resulting set of equations:
 first order systems 37

Okay, now let’s find our parameters. You find the following infor-
mation though observation, on the internet, and on the tags on your
water heater: Specific heat of water 4.2kJ/(kg K ), volume of water in
the heater 150L, density of water 1kg/L, power of the heater 4000W,
average temperature of water from the city 15o C, ambient temper-
ature in the room 21o C, thermal conductivity of the tank insulation
0.004W/(m C ), and the tank is a cylinder with diameter 0.5m and
height 1.2m (measured from the outside). The insulation is 5cm thick.
Find R and C. (remember to think carefully about assumptions!)

And finally, let’s put it all together to get our model: We’ll figure out the rest of the answer
(i.e., the ‘how long’ part of the question)
in the ‘System responses Part 2’ lecture.
38 engg*2400 engineering systems analysis

2.5.1 Assignment
1. From the text, do 11.1 (pg 390) (Hint 1: Don’t worry about capacitance. Hint 2: treat each plate as a
resistance, then write the equations and see if you can simplify it all down to one equation that satisfies
∆T = Req q, then Req is the equivalent resistance. Hint 3: Rules for combining resistors (like the kind
from circuits) that are all in parallel might help here)
2. From the text, do 11.5 a (pg 391)
3. From the text, do 11.9 a and c (pg 393) Hint: notice how there is no q variable... you’ll need to do some
rearranging to make the equations match! Also, FYI, these equations are structured in a state space form!
(We’ll get to state space later in the term)
4. Sketch the flow diagrams for the three problems above. Remember to treat each across variable (temper-
ature) as a node and remember to use conservation of flow.
 first order systems 39

2.6 Inputs and outputs

By the end of this section, you will be able to:

• Break a large, complicated problem up into smaller, simpler prob-

lems

• In your own words, explain the idea of using a ‘block’ to represent

a system

• In your own words, explain what an input and output is (in any
system)

We’ve been using the terms ‘inptut’ and ‘output’ a lot. In

this lecture, we’re going to take a closer look at these ideas to get a
deeper understanding of what they are and how to use them.
Inputs are signals that come from outside the system that the sys-
tem does not affect. Sometimes, we know and can control the input,
like the flow rate from the tap in our bathtub example. Thinking
about the hot water tanks from last day, what are the inputs (if any)?

Inputs (or sources) can supply either a through or an across variable.

What are these sources supplying? What kind of source was the water
tap in the bathtub? It was supplying a
known flow, so it is a ‘through’ source...
later, we’ll see that these are equivalent
to current sources in circuits.

Outputs are signals that we’re interested in. Yes, it is as

simple as that. Thinking about the hot water tank, what are the out-
puts (if any) and are they through or across variables?
40 engg*2400 engineering systems analysis

2.6.1 Breaking up large problems - identifying subsystems

Let’s see how we can use these ideas to help break a com-
plicated problem up into smaller parts. When heating milk
in an industrial setting, we don’t want to use an electric heater be-
cause it is too hot: the milk will scald and stick to the heater, affecting
the flavour. Instead, milk is heated using a two stage process: hot wa-
ter (possibly steam) is heated by an electric heater, then the hot water
is used to heat the milk. Note that modelling of thermal systems
is covered in much greater detail in
ENGG*3260.
Cold Milk inflow

Milk

Tm

Hot Water

Tw
Hot milk outflow

Cooled water inflow

Hot water outflow

qin

Identify the key components of the system (i.e., the sub-systems)

and those sub-system’s inputs and outputs, then draw a ‘block’ rep-
resenting each sub-system (much like we did with the car in the very
first lecture). For now, we’ll stick to modelling only
the thermal behaviour (i.e., we’re
going to ignore any issues around
determining fluid/mass flow rates,
etc.).
 first order systems 41

Notice how the systems interact with each-other: redraw the

blocks and connect up all of the signals that are the same (e.g., you
may have some inputs to one block that are outputs from another -
connect those lines!).

2.6.2 Modelling the milk tank

We now have all the info we need to model the whole

system. We’re going to do so by modelling one block at a time.
Let’s start with the milk tank. Assume that any milk removed from
the tank gets replaced by cold milk from the manufacturing plant’s There are a few important things to
point out about this equation: 1. T is
system, maintaining a constant volume in the tank. If we go back to
the absolute temperature of the fluid
our energy equation with this assumption, then this time we have that is flowing, not the temperature
of the fluid in the tank. 2. Here q
d mC p T dm represents heat flow, not volumetric
q(t) = = Cp T . flow rate... when discussing thermo,
dt dt
let’s use w for the volumetric flow rate.
Just like in our fluids example, let’s replace mass flow ṁ with the
volumetric flow rate Recall ρV = m and V̇ = w, so ṁ =
ρV̇ = ρw
q(t) = ṁ(t)C p T = ρC p Tw(t).

Great. Our other input is already well understood (we figured it out Note that an implicit assumption here is
when we did the heat exchanger!), so all that is left is to model the that the mass flow rate ṁ is constant, so
V̇ is also constant, as is w.
tank itself.
42 engg*2400 engineering systems analysis

Based on what we just did, can you write down the equation for
the hot water tank without doing any additional work? Hint: it is
essentially the same system as the milk tank, with one minor modifi-
cation.

The heat exchanger should also be easy - remember that we al-

ready modelled one back in the thermo lecture!

Excellent. Let’s collect all of our equations and our block diagram
in one place to ensure that we haven’t missed anything (remember,
we need to have enough equations that we can actually solve for our
unknowns!)
 first order systems 43

2.6.3 Assignment:
1. Answer the following questions as true or false. If you answer false, state a counter-example.
a) Inputs are always through variables
b) Outputs are always across variables
c) All of the across variables are outputs
d) All of the through variables are inputs
e) An output from one sub-system can not be an input to another sub-system
2. From the text, consider the problem outlined in 12.5 (pg 413). When we looked at this problem before,
we considered the flow w0 to be the output, but we didn’t give a reason. Consider the following sce-
narios and identify the appropriate outputs for each (note: listing every signal as an output is not a
reasonable answer!). Justify your selections.
a) Tank 2 is a water tower for the city of Guelph and Tank 1 manages overflow
b) This is an irrigation system
c) This is an industrial batch reactor system managing two connected, volume-sensitive, processes.
3. Consider the system given in Figure P12.7 in the text (p 414). Identify the sub-systems, their inputs and
outputs, and draw and label the interconnected block diagram. Write the input-output equation(s) for
each block.
4. Consider the system given in Figure P11.6 in the text (p 392). Let each of the ‘rooms’ be a sub-system of
interest. Draw and label the interconnected block diagram. Write the input-output equation(s) for each
block.
5. Consider the system given in Figure P12.5 in the text (p 413). Treat each tank and the pump as individ-
ual sub-systems. Draw and label the interconnected block diagram. Write the input-output equation(s)
for each block.
6. You want to model the coupled tanks system from the ENGG*4280 lab (shown on the next page), but it
is a complicated, interconnected problem.
a) What assumptions can/should you make to simplify this problem?
b) Identify the subsystems.
c) Identify the inputs and outputs from each subsystem.
d) Draw the complete, interconnected block diagram.
44 engg*2400 engineering systems analysis

Figure 2.1: The solid black lines repre-

sent pipes, γ is a ratio that expresses
the different flow rates through the pipe
branches (labeled out 1 and out 2), and
Doi (located at the bottom of each tank)
is the size of tank i’s outflow orifice.
This system is approximately three feet
tall, with each tank approximately 1
inch in diameter and 1 foot high. The
water basins are connected (i.e., effec-
tively, there is one communal basin)
and contain roughly ten times as much
water as can fit in the four tanks.
 first order systems 45

2.7 System responses - Part 2 Text: pages 200-201

By the end of this section, you should be able to: Secondary Text: Section 9.2

• Identify and express the step and impulse functions

• Derive the step and impulse responses for a first order linear ODE

• Determine k and τ (and the associated a and b) for a first order

system

• Explain, in your own words, what k and τ mean and how they
affect a first order system’s step and impulse response

• Explain, in your own words, the effect of initial conditions and

input size on a first order system’s response

Remember the water heater from a few lectures ago? Let’s use it
as the basis for this lecture. Recall that the problem was laid out as
follows:
You’ve just finished a shower and used up all of the hot water.
Your room-mate needs to know how long to wait until they can
shower. You’ve found the following information though observa-
tion, on the internet, and on the tags on your water heater: Specific
heat of water 4.2kJ/(kg K ), volume of water in the heater 150L, den-
sity of water 1kg/L, power of the heater 4000W, average temperature
of water from the city 15o C, ambient temperature in the room 21o C,
thermal conductivity of the tank insulation 0.004W/(m C ), and the
tank is a cylinder with diameter 0.5m and height 1.2m (measured
from the outside). The insulation is 5cm thick.
We already determined the relationship between the heater el-
ement and the temperature of the water in the tank in a previous
lecture:
1
C Ṫ (t) + ( T (t) − Ta ) = qin (t).
R

Yet again, we got a first order differential equation as

an answer to a problem. Maybe physics is trying to tell us some-
thing? Maybe it would be worth solving those equations in general?
If we’re going to get into math mode here, the first thing we should
do is recognize that both of those equations can be simplified and
written down in the following parametrized way: with x the input
and y the output, and k and τ (as yet unknown) parameters, we have

τ ẏ(t) + y(t) = kx (t).

46 engg*2400 engineering systems analysis

Example: Rearrange the hot water tank equation

1
C Ṫ (t) + ( T (t) − Ta ) = qin (t)
R
so that it fits the k and τ model above. Hint: You may want to con-
vert from absolute temperature T to a relative temperature Tr (t) =
T (t) − Ta

What are k and τ in terms of the tank parameters? We could do the same to the bathtub
problem from earlier in the term!

We can also write our differential equations in a slightly different

standard form (we move the 1 from the lowest order output to the
highest order output):

ẏ(t) + ay(t) = bx (t).

How do a and b relate to τ and k?

A very common problem in systems analysis looks like the fol-

lowing: given some (known) input (possibly zero) and some known
initial condition (possibly zero), determine the output. In our tem-
perature problem, that might be like asking ‘for a given initial tem-
perature of water in the tank and a known constant heat flow from
the heater, find the temperature of water in the tank as a function of
time’. To do so, we’d need to solve the differential equation... so let’s
do it!
 first order systems 47

2.7.1 No inputs - the Free Response

Before we worry too much about the effect of an input,

let’s try to understand what happens when there is no
input at all. The case where we have an initial condition and no
input is called the free response.
Let’s start by rewriting our differential equation with x = 0:

ẏ(t) + ay(t) = 0, y(0− ) = y0 ∈ R, t ≥ 0.

You should recognize this from Calculus as an Initial Value Problem

(IVP) whose solution satisfies

y(t) = e− at y0 = e−t/τ y0 , t ≥ 0.

I generated the figure below for a = 1/τ = 4 and for two initial
conditions: y0 = 2, −3. Notice that the title of the figure is
‘impulse response’ – this is because
I used a bit of a hack in Matlab to
generate this figure. We’ll see soon
how the impulse response and the free
response are similar to each other!

We can see that the initial condition y0 affects the starting location
and that the response always decays to zero. What affect does τ (or
the pole a) have? And what about k or b?
48 engg*2400 engineering systems analysis

2.7.2 Forced Response - introduction

Now let’s look at what happens when we have no initial

conditions, but we do have an input. This is called the forced
response.
As engineers, we’re usually interested in three main types of in-
puts: what happens when we turn something on (or off), what hap-
pens when we hit something really hard, and what happens when we
repeat the same process over and over. In math, these correspond to
three forcing functions: step, impulse, and sinusoid.

Step: The step input is zero for all negative time and 1 for all positive
time In this course, we give the step function the label u:
u(t)
(
0, t < 0
u(t) :=
1, t ≥ 0.
1
If we set the input to a system to be a step and initial conditions to
zero, then we call the corresponding output the step response.
t

Impulse: The impulse is infinity at zero and zero everywhere else,

but it is a special type of infinity... it satisfies
δ(t)
 
1
δ(t) := lim [u(t) − u(t − ε)] . 1/ε
ε →0 ε

Notice that the area under the curve is always exactly 1! This means
that Z t
δ(τ )dτ = u(t)! ε t
−∞
If we set the input to a system to be an impulse and the initial con-
dition to zero, then we call the corresponding output the impulse
response.

Sinusoid: A sinusoid can be a sine, a cosine, or an exponential (these

are complex sinusoids)9 . In this course, we’ll restrict ourselves to sine 9
Recall that et = cos(t) + j sin(t).
and cosine. If the input to the system is a sinusoid then the shape of
the output will depend on the frequency of the sinusoid. If we apply
inputs (one at a time) at all possible frequencies, then we refer to the
resulting collection of outputs as the frequency response.

For now, we’ll focus on steps and impulses. We’ll talk about fre-
quency responses more in a later lecture.
 first order systems 49

2.7.3 Impulse response of a first order system

Let’s consider the case of the impulse x ( t ) = δ ( t ) :

ẏ ( t ) + ay ( t ) = bδ ( t ) , y ( 0 − ) = 0, t ≥ 0.

Using Calculus, we get the following: If you want the details of this solution,
we can show you after class, or you can
k − t/τ ask your Calculus professor.
y im pul se ( t ) = be − at = e ,
τ
Okay, so we can see that b affects the starting location, what about
τ (or the pole a)? I generated the figure below for y 0 = 0, b = 4, and
a = 1/τ = 4.

How is this similar/different from the free response we looked at

earlier??
50 engg*2400 engineering systems analysis

Let’s look a little more closely at the impact of τ. It turns out Equivalently, we can say that the pole a
also determines how quickly the curve
that τ determines how quickly the curve moves and is called the
moves.
time constant. To see how τ (or a) affects things, I’ve plotted the case
where b = 1 and 1/τ = a = 2:

Recall that yimpulse (t) = be− at , and that b = 1 and a = 2. Using this
equation, find and label the points where t = τ = 0.5, t = 2τ = 1,
t = 3τ = 1.5, and t = 4τ = 2 on the curve.
Can we generalize this? Sure!

Scaling by b: On the curve above, sketch what will happen when we

have b = 0.5. In your own words, describe what is happening.

Scaling by a: On the curve above, sketch what will happen when we

have a = 4. In your own words, describe what is happening.
 first order systems 51

2.7.4 Step response of a first order system

Now we’ll turn to the case of the step input of some fixed
size x̂ ∈ R, so (
0, t < 0
x (t) = x̂u(t) =
x̂, t ≥ 0.
As engineers, we don’t care about what happened before we started
looking at the system (i.e., before time zero), so we can replace x
by x̂, use initial conditions, and rewrite our differential equation as
follows
ẏ(t) + ay(t) = b x̂, y(0− ) = 0, t ≥ 0.
Using the same techniques as before, we can find

b
1 − e− at x̂, t ≥ 0.


ystep (t) =
a
If we rewrite this using k and τ, we get
 
ystep (t) = k 1 − e−t/τ x̂, t ≥ 0.

Sketch the response for x̂ = 1, x̂ = 3, and x̂ = −3. Label where you

start and label the steady state (you may need to use k, a and/or b as
parameters in your sketch!)

Interesting... notice what’s happening with k? It is showing up in the

steady state value and acting as a gain10 on the input. 10
A gain is the ratio of the input to the
output. In this case, k is a gain on the
steady state and is given the name static
gain. In circuits, you’ll see this called
DC gain.
52 engg*2400 engineering systems analysis

Now we’d like to understand how the parameters k and τ (and

by extension a and b) affect the curve. Let’s look at the unit step
response (i.e., x̂ = 1)

y(t) = k (1 − e−t/τ ), t ≥ 0.

We already know (from our experimentation above) that the steady

state value will be k × x̂ , so what’s the role of τ? Very similar to the
impulse response! I’ve plotted the case of τ = 4 for you below. Using
what you learned from the impulse response, identify the value of y
for t = τ, t = 2τ, t = 3τ, and t = 4τ.
 first order systems 53

2.7.5 Complete response

It turns out that the solution to an Ordinary Differen-

tial Equation will always look like

Complete = forced + free.

Since we already understand both the forced and the free parts, we
don’t need to do any additional work to find the complete response!
Cool!

Example: Consider a differential equation 3ẏ(t) + y(t) = 2x (t) with

initial condition y(0− ) = −2. Sketch and write down equations for
the forced, free, and complete responses
54 engg*2400 engineering systems analysis

Putting it all together: How does this relate back to our heating
example?? Well, start by expressing τ and k in terms of R and C, then
sketch the response for the 4000W heater and the initial temperature
of 15o C. When does the water reach a temperature of 49o C? Q: What is the steady state tempera-
ture? Does this make sense? A: The
steady state temp is 21,200 (obviously
not possible). Hot water tanks would
have a temperature controller to ensure
the tank doesn’t overheat.
 first order systems 55

2.7.6 Assignment
1. Find k and τ (and a and b) for the bathtub example from section 2.1.
2. Consider the figure below containing 6 different responses for a first order system. You know that the
input is a step of size x̂ and the initial condition is given by y0 . Match the curve to the corresponding
pair of inputs and initial conditions.
a) x̂ = 1 and y0 = 2
b) x̂ = 5 and y0 = −2
c) x̂ = −5 and y0 = 2
d) x̂ = 1 and y0 = −2
e) x̂ = 5 and y0 = 2
f) x̂ = −5 and y0 = −2
g) Based on your answers, is it possible to determine k? If so, what is it?
h) (Challenging!) Find τ. Hint: use a combination of the graphs and the values you know/have calcu-
lated.

3. From the text, do 7.11 (note that the text uses u as a general input here, and not as the specific unit step
input). In all cases, indicate the forced and the free components of the solution. Hint: to answer c, you’ll
need to use superposition and the idea of time shifting... time shifting is an advanced idea that we
haven’t explicitly discussed, but the key idea is that, if you shift the input by a certain amount of time,
you shift the output by the same amount of time.
4. From the text, do 12.10 (p 416).
56 engg*2400 engineering systems analysis

5. Go back to the sketches that you created in the assignment from section 2.2 question 3 and relabel
the diagrams, this time including τ and 4τ (Hint: you weren’t given a differential equation, but you
should be able to look at those functions and relate them to a first order step response, then extract the
τ value). Use 4τ as the time at which the transients have ‘died out’ and the system is (approximately) at
steady state. Why do you think it makes sense to use 4τ? When/why might you use a different value
(e.g., 5τ or 3τ)?
6. You’re a coroner, and you come upon a body at 8am and you’d like to determine time of death. The
overnight temperature was 5o C, you know that the victim was approximately 70kg, that the specific heat
of the human body is approximately 4.2K J/(kg K ), and that the thermal resistance of fat and skin on a
person of this size is approximately 1/20 K/W.
a) Find a differential equation that relates the temperature of the body to the ambient temperature.
b) Find τ and k.
c) Standard (healthy) body temperature is approximately 37o C, and the temperature of the body when
you found it is 10o C. If you’re trying to determine time of death, what should you choose as your
initial condition and your forcing function?
d) Using your answers above, find the solution for the body temperature with respect to time and
sketch the associated curve. On your graph, label τ and the steady state.
e) Based on what you’ve done so far, determine the approximate time of death.
 first order systems 57

2.8 Electrical Systems and the Big Picture Text: Ch. 6

By the end of this section, you will be able to: Secondary Text: Chapter 4

• Identify the through and across variables and the fundamental

elements for electrical systems

• Apply resistive and capacitive element laws to determine current

and voltage change through and across electrical elements

• Find a differential equation that models a simple, first order elec-

trical system

• In your own words, explain the idea of an impedance

• Identify equivalencies across the three system types (Electrical,

Thermal, Hydraulic) Secondary Text: Section 2.4

• Express one system in terms of another (e.g., given a thermal

system, draw the equivalent circuit)

• Find H ( D ) for a system

Consider a very simple electrical system comprised of

voltage source, a resistor, and a capacitor: Note that analysis of electrical systems
is covered in much greater detail in
ENGG*2450.

iin (t) R C vc (t)

As usual, we’d like to model this system.

Signals of interest:
• The current i (t), usually in Amps (A). Current is the through variable.

• The Voltage e(t) (or sometimes v(t)) at various nodes, usually in Voltage is the across variable
Volts (V).

Node: A node is a connection point between elements. You can iden-

tify a node by looking for element connection points that all have the
same across variable. Through variables are conserved at a node.

Assumptions: Assume that all conductors are perfect and that all ele-
ments are ideal (e.g., a capacitor is purely capacitive with no resistive
component)
58 engg*2400 engineering systems analysis

We’re not going to worry about deriving the models for the resis-
tance R and the capacitance C, instead we’re going to assume (very
reasonably) that we can read them or measure them directly from
the element. That said, we still need the element laws for C and R in
electrical systems:

In this lecture, we’re going to focus more on putting together

some of the modelling ideas that we’ve been working on in previous
lectures. To that end, let’s start answering some questions about this
example system.

What should we use as our reference? Where should we put it? iin (t) R C vc (t)

What’s the input?

What’s the output? How do I express this voltage in terms of the

nodal voltages on either end of the element?

How many voltages are in this circuit? Is voltage conserved?

How many currents are in this circuit? Is current conserved?

 first order systems 59

Great. Now let’s find the differential equation relating the input
voltage to the output voltage. Start by sketching the flow diagram. Remember, when sketching flow
diagrams, connect elements at nodes,
and ensure that those nodes all have
the same across variable... labeling of
your initial diagram is very important
iin (t) R C vc (t) to ensure that your flow diagram is
correct.

What do you notice about the structure of your flow diagram com-
pared to the circuit? Okay... now write your equations and find the
differential equation relating the input to the output.

What’s τ? k? How long will it take for the transients in the capaci-
tor’s voltage to (approximately) die out?
60 engg*2400 engineering systems analysis

2.8.1 Impedance and the D operator

A major goal of this course is to find an overriding

structure to all of the different system types and their
solutions. So far, you should have noticed that all of the resistive
and capacitive elements have very similar element laws. We can go
a step further and say that there is one overarching type of element
law, and resistance and capacitance are just special cases. To do so,
we need the D operator:

dx (t) By the fundamental theorem of calcu-

= ẋ (t) = Dx (t).
dt lus, we can therefore write
Z t
1
x (t) = x (s)ds.
D 0

Example: Write the following differential equation using the D oper-

ator:
ẍ (t) + 12ẋ (t) = −4ẏ(t) + 0.1y(t).

Example: Write the following differential equation using the ‘dot’

notation:
( D2 − 2D + 12) x (t) = 12Dy(t).

With this in hand, notice all of our resistive laws become

∆across(t) = R through(t),

and our capacitive laws become

d ∆across(t)
through(t) = C = CD∆across(t).
dt
Now, here’s the dirty trick. We’re going to use the idea of an impedance.
If we define the impedance for a resistor to be Zr = R and the
impedance for a capacitor to be Zc = 1/(CD ), then we can write
all of our element laws as Note that impedances are functions of
the D operator!
∆across(t) = Z through(t).

It’s easy to see this works for a resistive element, so let’s double
check to see if it works for a capacitive one:
 first order systems 61

2.8.2 The Big Picture Part 1 - First Order

Now that we have impedances and we’ve modelled our

three first order system types, let’s put everything to-
gether in one place to see what we can leverage moving
forward.

All element laws satisfy:

Impedances satisfy:

ZR = Zc =

Conservation law for any element:

Across Through Resistive element Capacitive element Capacitor stores?

Hydraulic

Thermal

Electrical
62 engg*2400 engineering systems analysis

So, how do we use this to help with analysis? The table

above shows that all systems are equivalent (in some sense), so let’s
re-imagine the fluids problem from Lecture 2 as a circuit instead.
Recall that we already modelled the elements:

Now we just need to put them together in their flow diagram, then
convert that to a circuit: Why do we want to convert systems to
circuits? Because we’re going to learn
a tool called nodal analysis that will let
us write down our model equations
quickly and efficiently, and it is easiest
to apply nodal to a flow diagram,
which is essentially the same as a
circuit diagram!
 first order systems 63

2.8.3 H (D)

One last fun fact before we finish this big picture unit.
From our last example, notice that we can express our differential
equation with the D operator:

RP(t) + CDP(t) = qin (t)

If we factor and collect like terms on the LHS, we get

(CD + R) P(t) = qin (t),

and a bit of rearranging yields

1
P(t) = q ( t ).
(CD + R) in
| {z }
H (D)

It turns out that, for any of the systems we investigate in this

course, we can find an H ( D ) satisfying

Output = H ( D )Input.

This function H ( D ) is so important that there are many Matlab

functions created around it. For example, if you know H ( D ), you can
get the step response and the impulse response very easily.
For example, the following code will generate H ( D ) = 5D3+2 , then
plot the step and impulse responses. Note that tf is a command that is used
to make D a variable place-holder
D=tf(’s’); instead of a parameter – we’ll explain
more about this command near the end
H=3/(5*D+2)
of the term
step(H)
impulse(H)

I’ve placed the Matlab code that I used to generate the Responses
part 2 figures in Courselink for you to play with. You’ll note that it
leverages all of these functions.
For first order systems, we can also get the complete response
very easily (higher order systems are a bit more convoluted and need
state space tools – we’ll see how to handle this later). In the example
above, we can rearrange to find that a = 2/5 and b = 3/5. If the
initial condition is P(0− ) = 4, the input is sin(4t), and we want to
simulate from t = 0 to 15 at intervals of 0.01, we can get the complete
response using

a=2/5;
b=3/5;
sys=ss(-a,b,1,0);
64 engg*2400 engineering systems analysis

ic=4;

t=0:0.01:15;
in=sin(4*t);

lsim(sys,in,t,ic)

Note that the command lsim will plot the input and the output for
you on the same plot (nice!)
 first order systems 65

2.8.4 Summary of first order electrical systems

Assumptions:

• Ideal conductors

• Ideal components

Variables: The through variable is the current i and the across vari-
able is the voltage e.

Element Laws

• Resistance: (e+ t) − e− (t)) = i (t) R

• Capacitance: i (t) = C (ė+ (t) − ė− (t))

• Inductance: TBD

Sources:

• Current (provides known current i)

• Voltage (provides known voltage increase ∆e)

66 engg*2400 engineering systems analysis

2.8.5 Assignment:
1. From the text (p178), do 6.1, 6.4, and 6.17 (Hint: 6.17 this is a static system, so you don’t need deriva-
tives or D anywhere. Also, using a matrix to solve might come in handy here...). If your answer is first
order, identify k and τ and sketch the step response. Express your answers using the D operator.
2. From the text, do 7.7 a. and b. (pg 237). ‘No stored energy’ means that initial conditions are all zero.
du(t)
Hint 1: Notice that dt = δ(t). Hint 2: You may want to use superposition (i.e., the property that you
can solve for at one part of an input, then solve for another part of the input, and sum up the answers
to find the actual solution). Hint 3: Yes, you actually have everything you need to be able to do this
question without needing to use first principles calculus.
3. From the text, do 7.9 a. and b. (also, what is the time constant? what are the forced and free compo-
nents of the solution?)
4. For the thermal problem given in lecture 4, find the equivalent circuit.
5. For the hydraulic problem given in tutorial 2, find the equivalent circuit.
6. For the thermal problem given in figure P11.9 (p393), draw the flow diagram, the equivalent circuit, and
the equivalent hydraulic system.
7. This is a summative problem that takes many of the elements of modelling and response analysis
that you’ve learned so far and puts them together.
This is a modified version of a question from a ENGG*3430 Heat and Mass midterm (Yes, seriously!
This would be a mid/hard question for 2400 and an easy/fundamentals question in 3430).

Heater

Slab A Slab B

A square heater with surface 15cm × 15cm is inserted between two very tall slabs. Slab A is 2cm thick
(α = 50W/mC) and slab B is 1cm thick (α = 0.2W/mC). The convection coefficient on sides A and B
are 200W/m2 C and 50W/m2 C respectively. The temperature of the surrounding air is 25C. The heater is
rated at 1kW.
a) Assume that the system is at steady state (i.e., the thermal capacitance of the slabs can be neglected).
Find
i) The hottest temperature in the system.
ii) The outer surface temperature of the two slabs.
b) Now include the capacitance (this part was not on the 3430 midterm, but is totally reasonable for a
2400 midterm!)
i) What are the inputs and outputs of the system?
 first order systems 67

ii) Draw a flow diagram for the system.

iii) Find the equivalent circuit for the system.
iv) Find a differential equation that models the relationship between these input(s) and output(s).
v) How do you think your answers to a) above will change now that the capacitance has been
included?
3
Interesting things we can do with H ( D )

Now that we’ve finished with first order systems, there are
a few things we can learn to make our lives easier before we formally
move to second order and higher.

3.1 Nodal analysis Text: Section 6.4

By the end of this section, you will be able to:

• Explain the idea of nodal analysis in your own words

• Use nodal analysis to ‘solve’ a flow diagram (and therefore find a

system model)

• Write the nodal equations in matrix form and solve the resulting
matrix

Now that we have a decent handle on flow diagrams, the

D operator, and the idea of impedance, we can discuss a
valuable analysis tool called nodal analysis. The goal here
is to completely characterize the system behaviour in a systematic
way via a special function H ( D ) that’s going to come up a lot in this
course. To that end, consider the differential equation

τ ẏ ( t ) + y ( t ) = kx ( t ) .

If we rewrite this using our D notation, we get

τ Dy ( t ) + y ( t ) = kx ( t ) ,

but, if you recall from last lecture, we can ‘factor’ these D terms:

( τ D + 1 ) y ( t ) = kx ( t ) .
70 engg*2400 engineering systems analysis

We’d like to express our differential equation as follows:

output ( t ) = H ( D ) × input ( t ) ,

so, for the specific example above, y is the output, x is the input,
what’s H ( D ) ? This function is really important in
systems. The textbook talks about the
transfer function H (s), which is closely
related to H ( D ). For now, whenever
you see ‘transfer function’ in the text,
just replace the s with a D and move
on.

This function H ( D ) is one, nicely packaged function that lets us

relate the input to the output through differential operators. Wonder-
ful. It’s pretty, elegant, and nice math, but why do we care? Well, as
we progress through the rest of the course, this function is going to
keep coming up, and we’re going to be able to use it to predict most
of our system behaviours quickly and easily.
If H ( D ) is so useful, it would be great if we could have a quick
and easy way of analysing systems to generate it. So far, we have
been finding system models by combining conservation laws and
element laws, but this is very tedious. Nodal analysis works by writing
flow conservation equations in terms of the across variables instead
of the through variables (via the element law), generating a solvable
system of equations.

Nodal analysis basics:

1. Label all of the nodal across variables

2. Write the conservation equation in terms of a sum of across/impedance

instead of sum of through

3. (Optional:) To help reduce sign errors, if you are working on node

x, write your equations in the following way:

A x − A1 A x − A2
+ + · · · = 0;
Z1 Z2

i.e., always work from the node outwards. This means that you
have assumed that all flows are leaving the node.
 interesting things we can do with h ( d ) 71

Example: Let’s see how this procedure works on an example circuit.

Write the nodal equation at the node connection Z1 , Z2 , and Z3 :

Z3 Z4

Z5
Z2
Z1

ii1 (t) vi ( t ) ii2 (t)

We must be careful when handling sources! It doesn’t make

sense to talk about the impedance of an source, so we can’t write the
‘through’ of a source in it’s across/impedance form.
Dealing with through sources is easy: since we’re working with
conservation of flow, if there is a through source attached to the node
that we are working on, then we simply write the flow generated by
that source. Let’s do that for our example flow diagram.

The hard part is across sources. We have two cases:

• The source is connected to ground. Here, we just apply the defi-

nition of nodal voltages to determine that the voltage on the non-
grounded terminal is defined by only the across source and we do
not write a nodal equation at that node.

• The source is not connected to ground. We won’t analyze systems

with this structure in ENGG 2400, but you will in ENGG 2450.

Our end goal is to find the value of every nodal across variable.
Before proceeding, let’s think about how many equations we will
need.
72 engg*2400 engineering systems analysis

Okay, now let’s write down a systematic procedure, then try it to

see how it works.

1. Identify and label all of the nodes including the ground node.
Label any nodes attached to a grounded voltage source with the
(appropriately signed) value of that voltage source.

2. Choose a node. Imagine that all currents are leaving that node
and write KCL in voltage/impedance form at that node.

3. Repeat the previous step for all nodes except for the ground node
and any nodes attached to a voltage source.

4. For nodes attached to voltage sources (other than ground) follow

the procedure given above.

Z3 Z4

Z5
Z2
Z1

ii1 (t) vi ( t ) ii2 (t)

Now, combine everything we know to write the solution to the

circuit in matrix form.
 interesting things we can do with h(d) 73

Great. Now we can solve for every across variable in the system.
How can we use this information to find the through variables? Or a
system model?

Example: In the following flow diagram, use nodal analysis to find

the differential equation that relates Aout to Ain . Let Z2 and Z3 be
capacitors with with capacitance 2µF and Z1 be a resistor with resis-
tance 3Ω . Hint: set the bottom node to be ground.
Express your answer using D notation and find H ( D ) for this
system. Once you have done that, find a model relating the input Ain
to the current through the Z3 element.

Z1

Ain Z2 Z3 Aout
74 engg*2400 engineering systems analysis

3.1.1 Assignment:
1. Use nodal analysis and linear algebra to solve for all nodal voltages in the following figures from the
text (starting on pg177) P6.1, P6.4, P6.15, P6.16, and P6.17 (note: for P6.17, swap the location of the volt-
age source and the 3Ω resistor)
2. For the problem we solved at the end of the lecture, answer the following questions: (Hint: You should
be able to answer all of those questions by looking at H ( D ))
a) What is τ?
b) What is k?
c) If the input is a step, how long will it take for the system to reach steady state?
d) If there is no input, how long will it take for the initial voltage to die away?
e) Sketch the response of the system for zero initial conditions when the input is Ain (t) = u(t) − u(t −
∆t) when ∆t = τ, 2τ, 3τ, and 4τ. Which part of this response is a forced response? Which portion of
this response is similar to a free response? What’s happening to the capacitors?
f) Check your answers using Matlab. The code from the end of the Section 2.7 will be useful.
3. Consider the optional thermal problem from the end of tutorial 3. Sketch the flow diagram, write the
nodal equations in matrix form, and solve the resulting matrix to find an expression for each of the
temperatures as a function of the input heat flow. Find the associated H ( D ) for each temperature.
4. For Figure P12.7 (pg 414) in the text, draw the equivalent flow diagram and use nodal analysis to solve
for the pressures P1 and P2 as functions of the input pressure Pi . Find the associated H ( D ) for each
pressure.
 interesting things we can do with h ( d ) 75

3.2 Block Diagram Manipulation Text: Chapter 13

By the end of this section, you will be able to: Secondary Text: None

• Use basic block diagram structures to find the overall transfer

function(s) for a complicated, interconnected system.

• Use ‘fundamental’ blocks and the idea of state space to find a

block diagram representation for a given system.

Last day, we discussed representing complicated systems

as block diagrams, but (aside from the use of getting a
visual representation of the important interconnections)
we didn’t discuss how these diagrams can help us ‘solve’
the system. We’ll do that today We’ll also look at some other
ways that we can use block diagrams to deepen our understanding of
the relationship between state space and transfer function representa-
tions.

3.2.1 Simplifying block diagrams

A few lectures ago, we found the following block dia-

gram for a milk heater.

Tw

qin Tm
qe
Water Exch. Milk

qwc qcm

What we’re going to do now is identify five key structures that

will allow us to quickly and easily find the overall system transfer
function(s) for most interconnected systems. Below are the two fun- Note that the milk heater is actually a
damental blocks that we’ll need definitions for: nasty problem because it has interacting
feedback loops. We’ll solve it at the end
of this lecture
a
Note that G1 is assumed to be an H ( D )
type function!
b d
r y
G1 c
76 engg*2400 engineering systems analysis

Example: Find the equivalent H ( D ) the belongs in the empty box.

Hint: you may need to define some placeholder variables to help you
determine the correct answer. We just spent a while figuring out
how to go from a block diagram to an
y equivalent H ( D ), but can we go the
r r y
G1 G2 other way? Yes, and if you really want
to know how, you can learn from the
text.

r y
G1
r y

G2

y y
r r
G

r y r y
G

x x

r y
G
r y

H
 interesting things we can do with h ( d ) 77

So how do we tackle things with more than one input?

One at a time, than add them up!

Example: The block diagram below shows a control system for a

radio telescope antenna (described by H ( D )). The input n denotes
a wind gust disturbance that acts on the antenna, the input r is the
desired orientation of the antenna, and the output y is the actual
orientation of the antenna. C ( D ) represents the control engineer’s
attempt to determine the correct motor setting to ensure that the
antenna tracks the desired position. G ( D ) represents an attempt
by the control engineer to mitigate the effect of the wind. Find the
overall output as a function of the two inputs.

d
G

r y
C H
78 engg*2400 engineering systems analysis

Example: Now let’s tackle that milk problem! We want to find the
overall outputs as functions of the two inputs. For now, let’s try to
get Tm as a function of q in . You will do the rest in the homework!

Tw

qin Tm
qe
Water Exch. Milk

qwc qcm

First, we need to turn these blocks into things that look more like
what we’ve been dealing with during this lecture. Let’s go back to
the equations we developed to do so:

Rw Rqw
Water heater: Tw = (q + qwc )
Rw Rq wCw D + ( Rw + Rqw ) in
| {z }
=:Hw ( D )

Milk heater: Tm = Hm ( D )(qe + qcm )

Heat exchanger: qe = R1e ( Tw − Tm ) and

 
1 1
qwc = + Tw + R1e Tm
Rqw Re
| {z }
=:k
 interesting things we can do with h ( d ) 79

Now you should have something that looks like the following. This
is everything you need to solve, but doesn’t quite look like it. Let’s
manipulate it a bit more so that we can more easily see some of the
structures from earlier today.

qcm

qin Tw qe Tm
Hw 1/Re Hw
−

k Milk heater
Tm = Hm (qe + qcm )

qwc
1/Re

Water heater Heat Exchanger

Tw = Hw (qin + qwc ) qwc = kTw + R1e Tm
qe = R1e ( Tw − Tm )
80 engg*2400 engineering systems analysis

3.2.2 Assignment:
1. From the text starting on pg 448, do 13.1-3, 13.9, 13.14a, 13.16a-b.
2. Finish the milk problem!
 interesting things we can do with h ( d ) 81

3.3 Frequency response Text: Section 8.4

By the end of this section, you will be able to:

• Calculate and plot (Bode) the frequency response for a first order
system Secondary Text: Section 8.3.3

• Use a frequency response (numeric or figure) to determine steady

state behaviour of a first order system

A common forcing function in electrical systems is the

sinusoid (Alternating Current!), but many other systems also
have sinusoidal (or periodic) inputs. We’ve already taken the time to
understand how a first order system responds to a step or impulse,
so let’s take a close look at sinusoids. In particular, today we’re going
to see how H ( D ) relates to the frequency response. But to do that, I
need to show you just what the frequency response is, so let’s do that
now.
We’ve already discussed the response to a step and an impulse,
so let’s look at an input sinusoid sin ( ωt ) . To that end, consider the
differential equation

τ ẏ ( t ) + y ( t ) = k sin ( ωt ) .

After a lot of heavy Calculus, we get Ask your calc prof if you’re interested
in the details, the derivation is about a
k h   i page long and is beyond the scope of
y(t) = √ sin ωt − Tan− 1 ( τ ω ) − ωe − t/τ , t ≥ 0. 2400, but well within the scope of your
τ2 ω2 + 1 calculus course.

That’s pretty messy. Notice that the e t/τ term has shown up again,
but we already understand that behaviour (from the step and im-
pulse response discussion), so let’s focus on the steady state part:

k  
y(t) = √ sin ωt − Tan− 1 ( τ ω ) , t ≥ 0.
τ2 ω2 + 1

Example: Recall that the input was sin ( ωt ) . When we compare this
steady state response to that input, we can identify a phase offset and
gain (in the amplitude). What are they functions of?
82 engg*2400 engineering systems analysis

Example: For the case where τ = 10, ω = 0.1, and k = 4, find the
gain and the phase offset.

Example: Now let’s transfer these numbers to the diagram below.

a) On the figure, identify the effect of the initial conditions and

where (approximately) we start to see steady state behaviour

b) Identify the frequency of the input and the output. Are they the
same or different?

c) Identify the gain and the phase offset.

 interesting things we can do with h(d) 83

So the output is a function of the system parameters k and τ AND

of the input frequency ω. It would be nice if there was a structure
that we could leverage, especially as we go from first order to higher
order systems. It turns out that there is such a structure, and it comes
back to H ( D ) . Recall that (for first order) we got

k
H(D) = .
τD + 1
Let’s write this as a complex number instead, by replacing D by jω,
then convert to polar form and see what happens

k
H ( jω ) = = r ( jω ) e jθ ( jw ) .
τ ( jw ) + 1
Recall that, by definition, r = | H ( jw)|
and θ = ∠ H ( jw).

Well. That’s pretty cool. Is this true in general? Yes!!! Fantastic!

Now we have a nice way to calculate the gain and phase offset for
any sinusoidal input. In fact, it is so nice that we call this the fre-
quency response. It would be even nicer if we didn’t have to work our
way through complex numbers to get the answer... We can do that by
using the Bode Plot. This is a graphical representation of each of the
functions that we just calculated!

Example: For a first order system with k = 4 and τ = 10, find H ( jω )

for ω = 0.01 = 1/(10τ ), ω = 0.1 = 1/τ, and ω = 1 = 10/τ.
You could keep calculating for different choices of ω, and eventually
you’d be able to sketch the plot on the following page (notice that ω
is drawn on a logarithmic scale!)
84 engg*2400 engineering systems analysis

Now, with this figure, we can easily answer the question: what is the steady state output for this system
when the input is sin(0.06t)?

So we’ve done some math and drawn (and used) a figure. We’d like to understand what they actually
tell us...
 interesting things we can do with h(d) 85

I plotted the input/output pairs for the three choices of τ given

above. Things worth paying attention to include:

1. The transient behaviour (most noticeable in the ω = 1 case)

2. All inputs (blue) have the same amplitude

3. The outputs (red) change amplitude depending on the ω, and that

amplitude is consistent with the equations and with the Bode plot.

4. If you look at the second figure (I zoomed in a bit), you can see
the phase offset in steady state part of the responses, and that
phase offset is consistent with the equations and with the Bode
plot.
 86 engg*2400 engineering systems analysis

Now let’s look at an actual system to see why all of this

matters. Consider the circuit shown below. Find τ and k as func-
tions of R and C and sketch the Bode plot.

Vin C Vout

This circuit can be used as a filter that ‘passes’ all frequencies be-
low 1/RC and ‘attenuates’ all frequencies above 1/RC! If we let Be careful with your units!!! Hint:
1Hz = 2πrad/s.
R = 10kΩ and C = 15nF, find the cut-off frequency in kHz

So what does this mean? The circuit is an audio filter that keeps the
bass and removes the treble!
 interesting things we can do with h(d) 87

3.3.1 Assignment:
1. At the end of the lecture in section 3.1, we analysed a circuit, then answered a number of questions
about it in the homework. Now that you know about frequency response, answer one last question:
if the input is a sinusoid, at what frequency will the output be half the size of the input? Check your
answer in Matlab by using lsim (remember to simulate for long enough that you are looking at the
steady state part!)
2. For the circuit that we analyzed in the last lecture, find H ( D ) and sketch the Bode plot if R = 8200Ω
and C = 2.2µF.
3. Imagine that we are interested in a complicated circuit that satisfies the following:
1
H (D) = ,
D2 + 100D + 1
which yields the Bode Plot below. (note: you don’t have to know how to sketch these more complicated
plots, although you do need to know what the first order Bode plot looks like!)

Bode Diagram

0.8
Magnitude (abs)

0.6

0.4

0.01
0.2

0
0
Phase (deg)

-45

-90
-4 -3 -2 -1 0
10 10 10 10 10
Frequency (rad/s)

a) Let vin (t) = cos(0.01t).

i) Find the steady state output. Do this by using math and confirm your answer by reading the
Bode plot.
ii) Sketch the input and the output waveform on the same graph using the figure on the following
page as the input.
b) Let vin (t) = 2cos(1t).
i) Find the steady state output. Do this by using math and confirm your answer by reading the
Bode plot.
ii) Sketch the input and the output waveform on the same graph using the figure on the following
page as the input.
88 engg*2400 engineering systems analysis

1.5

0.5

-0.5

-1

-1.5

-2
0 2 4 6 8 10 12 14

0.8

0.6

0.4

0.2

-0.2

-0.4

-0.6

-0.8

-1
0 200 400 600 800 1000 1200 1400

4. Consider the system with

D−1
H (D) = √ .
( D + 3)( D + 1)
Find the steady state value of the output y if the input is x (t) = 3 sin(t). (Note, you can do this using
special triangles without your calculator!)
 interesting things we can do with h(d) 89

5. Calculate the magnitude, M, and phase angle, θ (radians), of the frequency response for the following
systems for ω = 0, 1, 2, 4, 8 radians/s

H (D) ω=0 ω=1 ω=2 ω=4 ω=8

1
H (D) = M = 0.25 M = 0.2425 M = 0.2236 M = 0.1768 M = 0.1118
D+4
θ=0 θ = −0.2450 θ = −0.4636 θ = −0.7854 θ = −1.1071

D+1
H (D) = M = 0.25 M = 0.3162 M = 0.3953 M = 0.4610 M = 0.4888
2D + 4
θ=0 θ = 0.3218 θ = 0.3218 θ = 0.2187 θ = 0.1206

1
H (D) = M = 0.25 M = 0.3162 M = 0.5000 M = 0.0791 M = 0.0165
D2 + D + 4
θ=0 θ = −0.3218 θ = −1.5708 θ = −2.8198 θ = −3.0090

2D + 3
H (D) = M=3 M = 0.7211 M = 0.4789 M = 0.3418 M = 0.2181
D2 + 5D + 1
θ=0 θ = −0.9828 θ = −0.9350 θ = −1.0023 θ = −1.1904

D
H (D) = M=0 M = 0.1667 M = 0.1617 M = 0.1413 M = 0.1010
D2 + 6D + 1
θ=0 θ=0 θ = −0.2450 θ = −0.5586 θ = −0.9197

D+4
H (D) = M=4 M = 1.4577 M = 0.4000 M = 0.0807 M = 0.0171
D3 + 3D2 + 3D + 1
θ=0 θ = −2.1112 θ = −2.8578 θ = 3.0911 θ = 3.0510
4
Second order systems

So far, we’ve spent a lot of time looking at first order

systems. We’ve also looked at systems that are composed of multi-
ple first order systems stacked together (e.g., two tanks and multiple
pipes). Some of these stacked systems are actually second order, but
we’ve implicitly limited ourselves to a certain subset of those second
order systems without even realizing it. We’re going to start by look-
ing at mechanical systems to better understand this fundamentally
different second order behaviour, then go back to our other system
types to see what we missed when we restricted ourselves to looking
only at resistive and capacitive elements.

4.1 Mechanical systems and Free Body Diagrams Text: Ch 2

By the end of this section, you will be able to: Secondary Text: Ch 3.3 and 3.6

• In your own words, explain the behaviour of a spring, a damper

(or viscous friction), and a mass.

• Write the physical laws that relate force to position, velocity, and
acceleration for a mass, spring, and damper (or viscous friction).

• Sketch a free body diagram for one mass mechanical systems.

• Use D’Alembert’s law to write the conservation law for mechanical

systems.

• Use the free body diagram and the conservation law to determine
the differential equation that models the behaviour of mechanical
systems.

Just like with our other system types, there are funda-
mental systems elements that we’re interested in when
92 engg*2400 engineering systems analysis

investigating mechanical systems. Unlike the other system

types, determining the through and across variables is not quite as
easy, so we’ll start by doing physics, and then see if we can figure out
what choice of through and across makes the most sense. For this
reason, we’re not going to refer to mechanical elements as resistive or
capacitive (at least, not yet).

Signals of interest:

• Forces F (t), usually in N. This is our through variable, although it

isn’t clear why yet.
• Displacement x (t), usually in m.

• Velocity v(t) = ẋ (t), usually in m/s. This is our across variable, although it
isn’t clear why yet.
• Acceleration a(t) = v̇(t) = ẍ (t), usually in m/s2 .

There are three main element types of interest: springs, masses,

and something called a damper (or a related idea called viscous
friction). We’ll start with the easiest one: springs.

4.1.1 Mechanical Elements

A spring stores potential energy: if you squish the spring, it

pushes back against you, and if you stretch the spring it will pull
back away from you. Various properties of the spring (e.g., metal
type, winding density, etc.) determine how strong that push or pull
will be, which can be captured via the spring constant K, which has
units N/m.

Just like with our other system types, we can use a reference.
When dealing with springs, it is easiest to choose the position refer-
ence such that x (t) = 0 when the spring is relaxed (or equivalently
unsprung). Using this reference, we can capture the behaviour with
Hook’s law:

Force felt by the hand(t) = F (t) = Kx (t).

Example: Let K = 10N/m. If the spring is at displacement x (t) = 5m

and is not in motion, what force is the spring applying to the hand?
 second order systems 93

What force is the hand applying to the spring? What force is the
spring applying to the wall? What force is the wall applying to the
spring? Hint: Think about Newton’s third
law... the one about equal and opposite
forces....

Okay, how do we handle it if the spring is not attached to a wall?

In the case shown below, how do we express the force felt by the
hand on the right?

x1 x2

Notice that what matters is the difference between x1 and x2 :

F (t) = K ( x2 (t) − x1 (t)).

This is an element law of sorts!

Dampers and viscous friction are different element, but

both resist the movement of an object. Viscous friction is
different than the static and dynamic friction that you learned about
in your mechanics course. Viscous friction occurs when there is a Static and dynamic friction depend
only on the mass of the object and the
fluid layer between two surfaces. As one surface moves, shear force
material properties!
is applied to the fluid, and the fluid then resists the motion of the
object. The faster the object moves, the higher the opposing force.
The properties of the fluid define a damping coefficient B, which
has units Ns/m. A damper (sometimes called a ‘dashpot’) leverages
viscous friction in an enclosed device and is typically an oil filled
cylinder with carefully placed holes and a moving piston. As the
94 engg*2400 engineering systems analysis

piston moves through the oil, the flow of the oil resists the movement
of the piston. The faster the piston moves, the higher the opposing
force. In both cases, the relationship between the applied force and The principle of direction of motion
versus force applied or felt is the same
the velocity is given by
as in the discussion of the spring.

F (t) = B( ẋ2 (t) − ẋ1 (t)) = B(v2 (t) − v1 (t))

v2

viscous
fluid

v1

Finally, we’ll look at a mass. You’ve probably heard in previ-

ous courses that masses have inertia, which relates to Newton’s first
law: an object in motion tends to stay in motion, and an object at
rest tends to stay at rest. His second law generalised this to allow a
mass’s motion to change, and states that the change in acceleration is
related to the applied forces (this is the conservation of force law!):

Inertia of mass = ma(t) = ∑ applied forces

4.1.2 Free Body Diagrams and D’Alembert’s law

You’ve seen free body diagrams (FBD) in previous physics,

mechanics, or highschool courses, but let’s look at them
again. The key idea here is that each mass or location where force
can be applied (sometime called a ‘massless point’) is treated as When we looked at springs earlier in
though it moves and acts independently of every other mass. We this lecture, each end of the spring was
a massless point!
draw the mass or location as a box, then label on the forces that are For now, we’re only looking at single
’felt’ by that mass, with appropriate directions. We use the direction mass systems!
of the position associated with that mass as the ‘positive’ direction.
 second order systems 95

Example: Let’s try drawing the FBD for the following simple Mass
Spring Damper (MBK) system: Ask yourself: which node is the refer-
ence in this case? I.e., at which node
x will we NOT write a conservation
equation?

f(t) M

Now, we’d like to use this FBD in the context of our course’s con-
servation laws and element laws, but right now we can’t quite do that
because the inertia term is ’hiding’. So, we’re going to slightly rear-
range Newton’s equation to more explicitly reflect the inertia term as
a (sort of) applied force of it’s own, and we’re going to include that
inertia in our FBD. This rearrangement is called
D’Alembert’s law, and allows us to
0= ∑ applied forces − ma (t) .
| {z }
treat a mass as a node and talk about
the conservation of the ‘flow’ of Force at
inertia that mass (or ‘node’).

Great, now let’s put this conservation law together with the FBD
to find the differential equation relating the applied force F to the
position of the mass x. Also find H ( D ). What’s the input and the output in this
case?
96 engg*2400 engineering systems analysis

4.1.3 FBD for systems with more than one mass

x1 x2

How do we handle systems with more than one mass? Ac-

tually, we’ve already made a good start when we looked at the spring
with two hands! The idea of holding one hand steady and figuring
out what’s happening to the other hand, then reversing the process,
then adding the results, is central to how we deal with more than one
mass.

Example: Draw the FBD(s) and write the associated equations.

x2
x3

F M2 B2
K2
K1

M1

B1

x1
 second order systems 97

4.1.4 Assignment:
Note that this is a long assignment! Many of them are somewhat repetitive and can be treated as drill to
help you build FBD skills.
1. Consider the system shown in figure P2.23 (p 45). Assume that the mass M1 is stationary. Find a dif-
ferential equation relating the output position x2 to the force due to gravity. Find the associated H ( D ).
(You can read about ideal pulleys on pg 31)
2. Consider the system showing in figure 2.17 in the text (pg30). Assume that the mass M1 is stationary.
Find a differential equation relating the output position x2 to the force due to gravity. Find the associ-
ated H ( D ).
3. From the text, do P2.26 (pg 46) Do not find Beq .
4. From the text, do P2.29 (pg47) (Hint: remember to draw a FBD at the massless point!).
5. From the text, do 2.23 (pg. 45). Compare your answer here to when you held M1 stationary in Q1. Does
this makes sense? Why/why not?
6. From the text chapter 2 starred problems (starting on page 39) do 2.1, 2.4, 2.7, 2.12, 2.15, 2.22.
7. (Challenge problems) From the text chapter 2 starred problems (starting on page 39) do 2.10, 2.18.
98 engg*2400 engineering systems analysis

4.2 Energy exchange, resonance, and inertiance Text: Ch 2

By the end of this section, you will be able to:

• Explain, in your own words, how a system can have an oscillatory

response when it is NOT excited by a sinusoid

• Explain, in your own words, the idea of resonance Secondary Text: pg243

• Identify the inertial elements in all system types and write down
the associated element law x

• Identify the across and through variables for mechanical systems K

f(t) M
Last day, we modelled a very simple mass-spring-damper
setup, and we talked about how the spring stores poten- B

tial energy and the mass stores kinetic energy. What’s

particularly interesting is that we can set up an energy exchange be-
tween these two elements. Consider the situation when the damping
coefficient B is zero (i.e., there is no damper). If I stretch the spring,
then let go, what happens to the position of the mass?

Okay, so we have a system that is bouncing (or oscillating) forever

(since we assumed zero friction). How does that happen? When I
initially stretch the spring, and hold the mass steady, where is the
energy held?

And when I release the mass, that energy gets (slowly) transferred to
where?

When does the mass have maximum kinetic energy and the spring
have zero potential?

Similarly, when does the mass have zero kinetic energy and the
spring have maximum potential? (Hint: there are TWO locations!)
 second order systems 99

Great, so we’ve established the idea that elements in a system

can exchange energy even though the elements are passive. We’ve
also identified that this energy exchange can lead to an oscillatory
behaviour in our system. How is this different than what we already
studied in our first order systems unit?

4.2.1 Resonance

Resonance is happens when we have a lossless energy ex-

change system (like our mass-spring above) that is excited with
exactly the right frequency: imagine if, instead of pulling then releas-
ing the mass, I excited it with a sinusoidal force, and that force was
chosen to push when the system was at maximum compression and
pull when the system was at maximum extension... it would have the
effect of making the oscillation grow and grow and grow to infinity!
Yikes! If this is true, we should be able to see it happen on the Bode
plot or when investigating H ( jω ), so let’s do that: Recall that we
found
1
H (D) = 2
,
MD + Bd + K
but B = 0, so this simplifies to

1
H (D) = .
MD2 + K
Find H ( jω ) and write it in polar form.

So, according to this result, the steady state of the position will be an
p
infinitely large sinusoid if the input looks like F (t) = sin( (K/M) t).
p
This is called resonance, and (in this case) (K/M ) is the resonant
frequency.
100 engg*2400 engineering systems analysis

Below, I’ve included the Bode plot for a system with M = 4kg and
K = 9N/m. Find and label the resonant frequency on the plot. What
does this plot tell you about the input/output behaviour above that
frequency? Below it?

4.2.2 Inertiance

We’ve noticed some interesting behaviour in our mechan-

ical systems around oscillation. Can our other system
types have the same behaviour? Did we miss something
when we were modelling them before? The short answer is yes
and yes :)
When we studied mechanical systems, we noticed that springs
store potential energy and masses store kinetic energy. In our other
systems, we also had elements that stored energy (by storing the
across variable). What were they?
 second order systems 101

How can we use this to relate mechanical systems to our other

three systems? Well, first notice that all of our capacitances can be
written as
1
∆across = through ⇔ CD∆across = through.
CD
Next, notice that we have ∆ type variables in our mechanical systems
as well:

So is that our across? And if so, what is our ‘capacitor’?

Hmm... that didn’t seem to work since dampers don’t actually store
energy. What happens if we consider ∆v to be our across instead?

Okay, so that implies that force is the through variable... does that
make sense?? Think about what happens when we apply a force
to one end of an element.... does the other end of the element ex-
perience the same force? Yes! Neat! So this means that force is our
through variable. We can now write our mechanical system equations
in terms of these across and through variables:

Mass:

Damper:

Spring:

Okay, what’s going on with the spring? Remember that the new
thing we noticed in our mechanical system was the ability to store
the force, or more broadly, the flow. It turns out that other systems
have elements that do this as well.

In hydraulic systems, inertiance is a way of modelling the ten-

dency of fluid in a pipe to want to keep moving (much like inertiance
in a solid mass). If the pipe is short, this effect is usually negligible,
102 engg*2400 engineering systems analysis

but for longer pipes this effect is crucial. Since this is an inertial ef-
fect, we’ll need to use F = ma, but we’ve talked before about how
fluids people don’t like to use mass, so lets replace it with density
and volume: m = ρV = ρA`1 . Additionally, we want to write our 1
Recall, ρ is density, V is volume
relationships in terms of the hydraulic across (P) and through (q), so
we’ll need to remember that force relates to pressure2 via f = AP 2
We’re working with hydraulics now,
so force is no longer the through
variable!

Using these relationships yields

A ( P1 − P2 ) = ρA` a = ρA` ẍ
| {z } |{z}
∆P m

Finally, we need to remember that our through variable in this

context is mass flow q, which is velocity through a surface area:
q(t) = Av(t) = A ẋ (t), which means that q̇(t) = A ẍ (t). Putting
this all together yields
ρ`
∆P = q̇.
A
|{z}
=:L
This is the element law for an inertiance, and can be expressed using Inertiance is the reason why the Bay
of Fundy experiences such large tidal
an impedance:
swings: the length of the bay matches
the volume (capacitance) of the bay so
closely that resonance occurs. Neat!

In electrical systems, we have a similar element, the induc-

tor. Recall from high school that charge moving through a coiled
wire generates a magnetic field - as you increase the flow, you in-
crease the strength of the magnetic field. If you then try to shut off
the current, the magnetic field slowly collapses, continuing to push
charge through the wire (just the same as the fluid kept being pushed
through the pipe above!). An inductor is such a coil, and ‘stores’ Notice that when the current is con-
flow via that magnetic field. Just like the hydraulic system, we can stant, so is the magnetic field, and the
inductor acts like a boring conductor, so
find the element law: the voltage drop is zero!

(V1 − V2 ) = LDi (t).

Thermal systems don’t have inertiance.

 second order systems 103

Example: (Time permitting! If we don’t do this in class, do it as part

of your homework!) Consider the following circuit. If the output is
the current through the inductor, find H ( D ) and identify if this is a
resonant system and, if so, what the resonant frequency is and which
elements are exchanging energy.

iin (t) 5nF 10mH

104 engg*2400 engineering systems analysis

4.2.3 Assignment:
2
1. A third year student models a complicated electrical system and finds that H ( D ) = 0.1D2 +12
. Find
H ( jω ) and write it in polar form. What is the resonant frequency of this system?
2. Redo the problem from Tutorial 2, but this time, also consider the inertiance in the long pipe. Find an
equation relating the pressure at the left side of that long pipe to the pump flow w p . Based on what
you found, do you think it is possible to have resonance in a fluid system? If so, what would need to be
true?
3. Consider the circuit below.

5nF

vin (t) 10mH

a) Find H ( D ) relating the input voltage to the voltage across the capacitor.
b) Let time be measured in micro-seconds. Find the steady state of the voltage drop across the capaci-
tor if the input is Vin (t) = sin(t) Volts.
c) For the same input, find the steady state of the current through the capacitor (hint: you know how to
integrate and differentiate sinusoids.... use that knowledge!)
d) Find the resonant frequency ωr of this system (include units in your answer).
e) What will happen if the input voltage is Vin (t) = sin(ωr t) Volts?
f) How are your answers the same/different from the example in class?
 second order systems 105

4.3 State Variables Text: Sections 3.1, 6.6

By the end of this section, you will be able to: Secondary Text: Section 2.2.3

• In your own words, explain the idea of state variables

• Identify the state variables for each element in each system type

• Express a system model in terms of its state variable equations

So far, we’ve focused on finding element laws, differ-

ential equations, and the function H ( D ) , but there is
another way to express our systems: State Space. This ap-
proach is particularly useful for higher order, complex systems. The
key idea here is to identify the inputs, the outputs, and the energy
storage elements and their associate variables, then write first order
differential equations in terms of only those key variables. The final
answer will be a set of n, first-order differential equations (one for
each energy storage variable, or state) plus one equation for each out-
put. But, this idea of ‘state’ or energy storage variable might muddy
the waters a bit, so let’s start by looking at a math example that has States are placeholder variables. Some-
times they have physical meaning, but
no relationship to any systems type problem.
the don’t always. You can think of them
as a variable that represents and inter-
Example: Consider the differential equation given by ÿ(t) + 5ẏ(t) + nal signal that we don’t care enough
about to label explicitly as an output
6y(t) = 3x (t) with input x and output y. The goal is to select some (although sometimes one or two of the
new variables z1 (t) and z2 (t) (these are called states so that we can states are chosen to be output(s)).
write an equivalent set of equations

ż1 (t) = f (z1 , z2 , x )

ż2 (t) = f (z1 , z2 , x )
y ( t ) = f ( z1 , z2 , x )

Let’s start by taking a naive approach: let z1 = y. If we do this, our

first equation is

That’s not useful yet, since we still have y on the RHS. What if we let
z2 = ẏ? Then that first equation becomes
106 engg*2400 engineering systems analysis

Okay, progress!! So what’s the second equation?

Are we stuck? It may look like it, but we’re not... we really just need
to go back to the original equation and back-substitute:

Now, what about that third equation?

Great! So we have our final state space representation, which we can

collect into matrix form:

ż = Az + Bx
y = Cz + Dx

So, does this mean we can always choose the first state to be the
output, then next state the derivative and so on? Unfortunately no.
Fortunately though, if we know something about the system, then
there are natural choices of states that will always work. The key is
to identify the energy storage elements and their associate variables, There are ‘tricks’ for going from any
differential equation to a state space
then write first order differential equations in terms of only those
representation, but you’ll learn more
variables. about those when you take ENGG*3410.
So, how do we choose the energy storage variables to be interested
in? Typically, we choose either the ∆across or the through variable
associated with that element, and we typically choose whichever
variable is ‘stored’. So, for a circuit, we’d choose the current through
the inductor and the voltage change across a capacitor.
 second order systems 107

Example: Consider the circuit shown below. Identify the energy

storage elements and their associated variables, then find the state
space representation. You can check your answers by using
nodal analysis and confirming that you
get an equivalent result! Fun fact: if
R1 R2
you find your state space representation
and put it in matrix form, you can get
the equivalent H ( D ) via the equation
vin (t) C L C (sI − A)−1 B + D!
108 engg*2400 engineering systems analysis

Given that thermal and hydraulic systems can be writ- x2

x3
ten as circuits, we can apply the same principle to them.
F M2 B2
But what about mechanical systems? We still haven’t formally K2
found the equivalency for them yet (we’ll do so in the ’big picture’ K1

lecture two lectures from now), but for now let’s see what we can fig- M1
ure out from what we already know. In an earlier lecture, we found
B1
the FBDs for a two mass system, and the resulting equations were
x1

Mass 1: 0 = M1 ẍ1 + B1 ẋ1 + K1 x1 + B2 ( ẋ1 − ẋ2 )

Mass 2: 0 = M2 ẍ2 + B2 ( ẋ2 − ẋ1 ) + K2 ( x2 − x3 )
Massless point: f = K2 ( x 2 − x 3 )

Before we go any further, we should think about the inputs, the

outputs, and how many states we’ll need.

We know that the spring stores potential energy and the masses
store kinetic, but which variables should we choose? If we were to
think like a circuit, we’d choose ∆v for the mass, which corresponds
exactly to ∆across. We’d also choose ‘force’ for the spring, but that
’force’ term doesn’t show up directly in our equations... instead,
we have k( xi − x j ) type terms, so instead of choosing the through
variable f , we’ll choose the equivalent, ∆x. We know from a moment
ago that we’ll need four states. Based on what we just discussed,
what should those four states be?

Okay, great. Now it just remains to do a bunch of algebra and

manipulation to get the final answer!
 second order systems 109

4.3.1 State-space and Matlab

Way back in Chapter 2 we learned about using Matlab to

simulate systems, but we said we’d need state space to do
the really interesting stuff. You can simulate any input and
any set of initial conditions using a combination fo the ss and lsim
commands.

Example: Consider the system given by the state-space representa-

tion
" # " #
0 1 0
ẋ (t) = x (t) + q(t)
−2 −3 1
| {z } | {z }
=A =B
h i
y(t) = 1 0 x ( t ).
| {z }
=C

Find the response when the input is 2 cos(4t) and the initial condi-
tions are x1 (0− ) = −1 and x2 (0− ) = 5.

A=[0 1; -2 -3]
B=[0;1]
C=[1 0]
D=0
sys=ss(A,B,C,D)
t=0:0.1:10;
q=2*cos(4*t);
ic=[-1;5]
lsim(sys,q,t,ic)

We can also switch between a transfer function and state space

and back again by using ss2tf and tf2ss! (Note that there are an infi-
nite number of possible state space representations for one transfer
function, so the output of the tf2ss command may not yield what you
expect!)
110 engg*2400 engineering systems analysis

4.3.2 Assignment:
1. Finish the mechanical problem given at the end of the lecture and express your answer in matrix form.
Recall that we use z1 = x1 , z2 = ẋ1 = v1 , z3 = x2 , z4 = ẋ2 . Treat x1 as the output. How does your
answer change if the input is x3 instead of F?
2. From the text, do 3.5, 3.9, 3.13, 3.18, and 3.29 (starts on p 70)
3. From the text, do 6.9, 6.21, and 6.38 (starts on p 179). Also do 6.3, 6.7, 6.12, and 6.18.
4. From the text, do 11.7, 11.9, and 11.11 (starts on p 392) For 11.9, find H ( D ) and compare to what you
got when you solved this system in a previous assignment.
5. From the text, do 12.5 (take wo and p̂2 as the outputs) and 12.7 a (also find H ( D )) (starts on p413)
 second order systems 111

4.4 Second order step and impulse responses Text: Subsection starting on p258

By the end of this section, you will be able to: Secondary Text: Sections E.3.2 and 8.4.1

• In your own words, explain the idea of damping and natural fre-
quency

• Identify whether a system is over, under, critically, or un-damped

• Explain what it means to be over, under, critically, or un-damped

• Sketch and label the step and impulse responses for underdamped
systems

So, now we know everything there is to know about all

of the different elements in all of the different system
types, and we also know all the different ways to model
them. We’ve even looked at frequency response and resonance. The
last thing to do before combining everything in the ‘Big Picture’ is to
look at step and impulse responses.
Just like with first order systems, we can write our second order
systems in standard form: with the input x (t) and output y(t), we
can write

kωn2
y(t) = x (t)
1D2+ 2ζωn D + ωn2
k
= 1 2ζ
x (t), ζ ≥ 0, ωn > 0.
D 2 + D + 1
w 2 ω n
n

Again, just like with first order systems when we investigated k and Just like before, k is the static gain. Our
new terms are ζ, the damping ratio and
τ, here we want to understand what k, ζ, and ωn mean.
ωn , the un-damped natural frequency.
Before jumping into the math, consider that we’ve already talked
about resonance, where we got something that looked like

1
H (D) = ,
MD2 +K
if we put this into standard form, we get

So, it looks like wn may relate somehow to resonance. At a very

minimum, we should expect the ωn term to have something to do
with the frequency of oscillation of the system.
112 engg*2400 engineering systems analysis

4.4.1 Characteristic Polynomial

We’ve been looking at this H ( D ) function for a while now,

and part of the reason for that is the denominator. It
turns out that the denominator of H ( D ) tells us very important
things about the response to any input, so much so that it has a
name: the characteristic polynomial. For example, we already know
about τ and first order systems... notice that 1/τ is the root of the
polynomial τD + 1 and is called the pole of the first order system.
Well, if we have a second order system, then the denominator is
1D2 + 2ζωn D + ωn2 and has two roots and therefore two poles. Just
like how τ determined how long it takes for the first order system
to ‘reach’ steady state, the poles of higher order systems determine
fundamental system behaviour too! So, let’s take a minute to find the
roots of our second order characteristic polynomial (i.e., let’s find the
poles). Hint: you’ll need the quadratic equation.

Okay, so what does this mean? We can have real or complex roots:

Real and distinct:

Real and repeated:

Purely imaginary:

Complex:

Sketch these locations on the complex plane below:

 second order systems 113

Now, with all this in hand, let’s rewrite the poles in their polar
form and sketch the result in the complex plan to see if we notice
anything interesting about what happens as ζ moves from 0 to 1. These ideas are crucial in many me-
chanical courses, such as Vibrations
and in any other system where energy
exchange or oscillation is present.

Does that matter? Yes, it turns out that it does! To see how, we
need to look at one system type at a time.

4.4.2 Real roots: over and critically damped

Let’s start with the simplest case: real roots. Let’s assume
that we can write the differential equation such that

k
, τ1 , τ2 ∈ R.
(τ1 D + 1)(τ2 D + 1)

Notice that we rewrite this in the following way3 3

This uses a technique called partial
fraction, which you’ll learn in calculus
τ1 τ2 !
in a week or two
k τ1 −τ2 −τ1 +τ2
=k + ,
(τ1 D + 1)(τ2 D + 1) τ1 D + 1 τ2 D + 1

which is just two first order systems added together! This is signif- If τ1 6= τ2 , in calculus you called this
icant, because it means that we can write the output as a sum of two ‘non-repeated roots’, but in systems
we call these systems overdamped. If
first order system responses τ1 = τ2 , in calculus you said this was
‘repeated roots’, but in systems we
kτ1 kτ2 call the system critically damped and
τ1 −τ2 −τ1 +τ2
y(t) = x (t) + x ( t ), the partial fraction expansion yields
τ1 D + 1 τ2 D + 1 something slightly different - you’ve
already investigate the details of both of
which means that the step response is the sum of those two first these types of differential equations in
order system responses: your calculus course.

kτ1 kτ2
ystep = (1 − et/τ1 ) + (1 − et/τ2 ), t ≥ 0.
τ1 − τ2 −τ1 + τ2
You’ll spend some time exploring this and the associated impulse
response in the assignment.
114 engg*2400 engineering systems analysis

4.4.3 Complex roots: under and un-damped

Complex poles make things harder, since we need to do

calculus again to get a result. Here we investigate the case Just like with frequency response, the
calculus here is too long and involved
where ζ ∈ [0, 1) and we are interested in the step response. Recall
to spend lecture time on, so we’ll skip
that the differential equation is given by straight to the answer. You DO need
to know how to do it for your calculus
k course.
y(t) = 2ζ
x ( t ).
1
w2n
D2 + ωn D +1

It turns out that the step response is given by

"  q #
1 − ζωn t − 1
k 1− p e sin ωn 1 − ζ 2 t + cos ζ , t≥0
1 − ζ2

and it looks like

Step Response

0.8

0.7

0.6

0.5
Amplitude

0.4

0.3

0.2

0.1

0
0 2 4 6 8 10 12 14 16
Time (seconds)

This is the step response for

Observe that the final value is the static gain k, just like it was in the 1
H (D) = .
first order case.... indeed, this is always true! Also, notice that the D2 + 0.7D + 2

value of ζωn (i.e. the real part of the pole) seems to be related to how Find k, wn , and ζ. Do these values make
sense to you?
quickly the curve decays, so it has a similar role to 1/τ from the first
order case. We also see that the frequency of oscillation depends on
the damped frequency
q
ωd = ωn 1 − ζ 2 ,

which is the imaginary part of the pole.

Another interesting effect of the sinusoidal action is that second
order systems can have overshoot, where first order systems can not.
We can even quantify this overshoot via the following equation:
√−ζπ
%OS = e 1− ζ 2 ,
 second order systems 115

which occurs at time

π
tmax = .
ωd
116 engg*2400 engineering systems analysis

Let’s take a second to look at two plots, one that shows what hap-
pens as we change ζ and the other as we change ωn :

Important intuition: The exponential

decays faster when ζωn gets bigger.
p
The oscillations get faster as ωn 1 − ζ 2
gets bigger. Together, this means that
the system gets faster as ωn gets bigger.
Additionally, the amount of overshoot
increases as ζ gets closer to zero.

Systems concepts question: If we put in a step of size 2 (instead of

1), what would the final value of the step response be? Would the
time at which it settles down and stops oscillating change?
 second order systems 117

Example: Let’s do a quick check against our mechanical system to

see if we can relate any of these concepts to a real system. Recall that
our simple mechanical system yielded

1
H (D) = .
MD2 + BD + K
Put this in standard form, find k, ζ, and ωn . Which parameters affect
which terms? Does this make sense based on what we know about A great way to gain intuition about the
behaviour of second order systems is
things like friction, resonance, and steady state? Why/why not?
via this Mass-Spring-damper system,
you’ll explore it further in Tutorial 7.

One last interesting tidbit before we move on to impulse: Here’s

the Bode plot for a constant ωn = 0.5 and changing ζ. Is this consis-
tent with what you’d expect based on the ideas of resonance that we
discussed a few lectures ago?
118 engg*2400 engineering systems analysis

The impulse response is given by Let’s sanity check this... it should be the
derivative of the step response.... yup
it is! (Note that this is not trivial to do!!
 q 
kω
p n e−ζωn t sin ωn 1 − ζ 2 t , t ≥ 0 You’ll need some trig identities.)
1 − ζ2

Step Response

0.6

0.5

0.4

0.3
Amplitude

0.2

0.1

-0.1

-0.2

-0.3
0 2 4 6 8 10 12 14 16 18
Time (seconds)

Again, notice that ζωn determines how quickly the curve decays
p This is the impulse response for the
and that the frequency of oscillation depends on ωn 1 − ζ 2 . As with transfer function
the first order case, the static gain does not explicitly show up in this 1
H (D) = .
figure. Also, notice that this impulse response differs from the first D2 + 0.7D + 2

order case in that the curve starts out at zero instead of at a constant
(non-zero) value.
As with the step responses, let’s take a look at two plots, one that
shows what happens as we change ζ and the other as we change ωn :
 second order systems 119

4.4.4 Summary of damping

Note that our standard form does not
Un-damped: ζ = 0 two purely imaginary (conjugate) poles include the case where there are two
Underdamped: ζ ∈ (0, 1) two complex conjugate poles real poles with one positive and one
negative - we didn’t explicitly discuss
Critically damped: ζ = 1 two repeated real poles this in this lecture, but we’ll come back
Overdamped: ζ ≥ 1 two non-repeated real poles to it later lecture.
Unstable: ζ < 0 poles have negative real part (may be real or com-
plex poles)
Important variables when plotting underdamped systems:
p
• Damped frequency ωd = ωn 1 − ζ 2
√−ζπ
• Percent Overshoot %OS = e 1− ζ 2

π
• Time to first peak tmax = ωd

• Similar role to τ: ζωn

120 engg*2400 engineering systems analysis

4.4.5 Assignment:
1
1. Consider the system given by H ( D ) = (τ1 D +1)(τ2 D +1)
with τ1 , τ2 ∈ R. Let τ1 = 1. Consider four cases:
• τ2 = 1
• τ2 = 2
• τ2 = 10
• τ2 = 50
a) Find an equation for, and then sketch the step responses.
b) For the step responses, calculate the value of the output, y, at time t = 4τ2 . What do you notice as τ2
changes?
c) For the case where the system is critically damped, calculate the value of the step response y at time
t = 5.8τ. What does this tell you about how this type of system relates to a ‘pure’ first order system?
d) Find an equation for, and then sketch the impulse responses.
e) For the impulse responses, calculate the value of the output, y, at time t = 4τ2 . What do you notice
as τ2 changes?
f) For the case where the system is critically damped, calculate the value of the impulse response y at
time t = 5.8τ. What does this tell you about how this type of system relates to a ‘pure’ first order
system?
g) Use Matlab to check all of your answers (you don’t need state-space!).
2. From the text (pg. 289-293), do 8.8, 8.9, 8.10, 8.12, 8.34, and 8.36.
3. From the text pg. 451, do 13.18b and d (recall that you already solved a in chapter 3!)
4. On the next page, you’ll find a favourite assignment problem from one of my old professors. The plots
on the bottom show the location of the poles as mapped into the complex plane; ignore the poles given
in (5) and (7) and their associated curves (b) and (d). Hint: From the pole locations given in the complex
plane, you can determine ζ and ωn or your value(s) of τ, whichever is appropriate. Hint2: You should
be able to figure out (6) by process of elimination. Based on what you discovered in Q1, does it make
sense that the addition of a real, slow pole has this affect on the under-damped step response?
second order systems 121
122 engg*2400 engineering systems analysis

4.5 Big Picture - Second order

By the end of this section, you will be able to:

• Explain the equivalencies between the four system types

• Express any system type as an equivalent circuit and vice versa

(mostly)

We’ve been holding off on this for a long time, but we’re
finally ready to do it: finding the equivalent circuit (or
flow diagram) for mechanical systems. The key idea here
is to leverage the equivalencies we determined in the table above to
translate the free body diagram into a flow diagram. Just like when
we did flow diagrams back in the first order unit, the key to writing
the flow diagram is to identify the nodes and to connect each node
to other nodes via elements. So, what are the nodes in a mechanical Hint: remember that we like to do
conservation of flow at a node and that
system?
flow in mechanical systems is force, so
where in mechanical systems do we
write conservation of force equations?

Example: Let’s try a very simple example:

F
M B

x1

Holy crap. That’s the same circuit we got in the First Order big
picture lecture!
 second order systems 123

Okay, now we’re finally ready to make our ‘Big picture’ table.
After this, we’ll do some examples.

All element laws satisfy:

Impedances satisfy:

ZR = Zc = Zl =

Hydraulic Thermal Electrical Mechanical

Across

Through

Resistance

Stores?

State Variable?

Capacitance

Stores?

State Variable?

Inductance

Stores?

State Variable?
124 engg*2400 engineering systems analysis

Example: Okay, let’s step it up a notch. What happens when we have

two masses? Find the equivalent circuit and fluid models.

K2 K1
F
M2 M1

B2 B1

x2 x1

Great. That’s second order, all done! Now we just need to learn
about Laplace and about interconnected systems, and the course is
over :)
 second order systems 125

4.5.1 Assignment:
1. For the problems given in 2.10, 2.23, 2.27 (starts p41), find the equivalent circuit.
2. For the problems given in 6.7 (p179), find the equivalent mechanical and fluid systems. Is there an
equivalent thermal system? Why/why not?
5
Laplace Transform

We’re finally here!! We’ve been hinting at this tool for

the entire term, and now we finally get to use it. The main
reason that we want to use Laplace is that it lets us turn calculus
problems into algebra problems, which is pretty cool. We’ll start This chapter weaves very closely with
your calculus course. Most of the
by going through some definitions and (finally!) learning about the
time we should be one or two lectures
transfer function, and then we’ll get into how to use Laplace to solve after calculus, but there may be the
differential equations that are of interest to us in this course. occasional concept that comes ahead.

5.1 Definitions and Transfer Functions Text: Sections 7.2 and 7.3

By the end of this section, you will be able to: Secondary Text: Section F.1-F.3

• Apply the definition of the Laplace transform to derive the trans-

form of a step and an exponential.

• Identify and apply the properties of Laplace transform to convert a

differential equation into the frequency domain.

• Identify the transfer function.

• Identify the free and forced part of a response when expressed in

frequency domain.

• In your own words, explain how initial conditions relate to energy

storage.

Let’s start by defining the Laplace transform and using

it to determine the transformation of some important
signals.
Given a function f : R → R,1 the Laplace Transform is defined via 1
This notation means that the function
f takes a real number as its input and
Z ∞
yields a real number as its output.
F (s) := L { f (t)} := f (t)e−st dt
0
128 engg*2400 engineering systems analysis

Note that the lower limit of integration is a zero!2 This means that 2
This is why we call the Laplace trans-
any part of the function f that occurs during t ≤ 0 will not affect the form a ‘single sided’ transform.

Laplace transform of that signal; therefore, when we take the inverse

of a Laplace transform, our answer is only valid for time greater
than zero!3 3
This is not the same as saying that our
Keep in mind that the variable s is a complex number (when nec- solution is zero for t ≤ 0; in fact, for
t ≤ 0 it is more accurate to say that we
essary, we’ll write it as s = σ + jω). We often refer to the Laplace don’t know what the solution is.
transform of a time domain (TD) signal as a frequency domain (FD) Notation: In this chapter, we will use
signal. upper case letters for FD signals and
lower case letters for TD signals.

Example: Find the Laplace transform of a unit step.

Example: Find the Laplace transform of f (t) = e− at , a∈R

 laplace transform 129

Rather than using first principles to find transforms of signals,

we typically use tables. In this course, the following signals will be
important:

Time Domain f (t) Laplace Transform F (s)

δ(t) 1

1
u(t) s

e− at 1
s+ a

te− at 1
( s + a )2

ω
sin(ωt) s2 + ω 2

s
cos(ωt) s2 + ω 2

e− at sin(ωt) ω
( s + a )2 + ω 2

e−at cos(ωt) s+ a
( s + a )2 + ω 2

We will also be leveraging some very important properties of the

Laplace transform:

L { f (t)} = F (s) Definition of notation

L { a f (t) + bg(t)} = aF (s) + bG (s) Linearity: superposition and homogeneity
L f˙(t) = sF (s) − f (0− )

Differentiation
L f¨(t) = s2 F (s) − s f (0− ) − f˙(0− )

Second order Differentiation
Z t 
1
L f (τ )dτ = F (s) Integration
0 s

5.1.1 Transfer functions and the free and forced responses

Okay, now that we have these definitions, let’s apply the

Laplace transform to one of our systems to see what hap-
pens. Consider the simple mass spring damper system. We know
130 engg*2400 engineering systems analysis

that we have three ways of expressing a model of the system:

x
1. Differential equation: M ẍ (t) + B ẋ (t) + Kx (t) = f (t)
1 K
2. H ( D ) : x (t) = f (t)
MD2 + BD + K
| {z } f(t) M
= H (D)

ż(t) = Az(t) + B f (t) B

3. State space:
x (t) = Cz(t) + D f (t)

We already know how differential equations and H ( D ) related to

each other, but state space is still a bit confusing. Let’s use Laplace
transform to see what else we can learn.

Example: Apply Laplace transform to the differential equation (item

1.), then isolate for the output X (s).

Things to notice: the free and forced

parts are explicit AND H (s) is associ-
ated with the forced part. Also notice
that the denominator (and therefore the
poles) is the same in both the forced
and the free part.

Important: We call H (s) the transfer function. Notice how it is just

H ( D ) with the D replaced by an s, but also notice that the equation
we have here is different from when we express the system with
H ( D ). Why is that?
 laplace transform 131

Okay, so how do we use these tools together? Well, let’s see

if we can figure out a step response. Consider the mechanical system
given above. Let M = 2, B = 1, and K = 10 in a coherent set of
units. Let the input force f (t) be a unit step and consider the initial
conditions x (0− ) = 4 and ẋ (0− ) = −1. Find the step response
expressed in frequency domain X (s).

Outstanding. Now we just need to know how to unwrap this and

turn it back into the time domain to find x (t). We’ll learn how in the
next lecture.
132 engg*2400 engineering systems analysis

5.1.2 Relationship to state space (time permitting)

Okay, can we use Laplace transform to help us learn

more about what’s going on with the state space repre-
sentation (item 3.)? Possibly, but first we actually need to choose
states and find the matrices ( A, B, C, D ). There are lots of acceptable
choices, but for now let’s choose the following:

z1 (t) = kx (t) The spring stores its force f k (t) = kx (t)

z2 (t) = v(t) = ẋ (t) The mass stores its velocity.

With this definition, you can find the following state space represen-
tation (I encourage you to try this at home!)

ż1 (t) = kz2 (t)

1 B 1
ż2 (t) = − z1 ( t ) − z2 ( t ) + f (t)
M M M
1
x (t) = z ( t ),
K 1
which has matrix form
" # " #
0 K 0
ż(t) = 1 B z(t) + 1 f (t)
−M −M M
| {z } | {z }
=A =B
h i
x (t) = 1 z ( t ) + [0] f ( t ),
K 0
| {z } |{z}
=D
=C

Example: Apply the Laplace transform, then isolate for Z (s). Write
your answer in matrix form and also in non-matrix form.

In the solution above, identify the free and the forced part for each
state equation. For the free part, what does the initial condition rep-
resent? How does this fit with concepts of energy storage, memory,
and dynamic systems?
 laplace transform 133

Example: Finally, substitute into the general form (the one using
( A, B, C, D )) using your answer above to solve for X (s). Identify the
free and forced part. Can you find the transfer function here? Finish
the problem by substituting in for the actual values of the matrices
and confirm that your answer matches the solution we got when we
took the transform of the differential equation (in item 1.)

Next day, we’ll look at how we can use this new form of the differ-
ential equation to solve for many different forcing functions.
134 engg*2400 engineering systems analysis

5.1.3 Assignment:
1. From the text, do 7.3 (p236)
2. From the text, consider the differential equations given in 7.7, 7.8, and 7.11 (starts on p 237). Note: For
7.7 and 7.8, the input is i and the output is e. For each differential equation,
a) Apply the Laplace transform to the differential equations
b) Find the Forced and Free part of the responses
c) Find the Transfer function
d) Find the poles
e) Zeros are defined to be the roots of the numerator of the transfer function. Find the zeros. Note that
a zero at 0 is NOT the same as having no zeros at all.
 laplace transform 135

5.2 Partial Fraction Expansion Text: Section 7.6

By the end of this section, you will be able to: Secondary Text: F.7

• Use Partial Fraction Expansion (PFE) to convert Frequency Do-

main (Laplace) signals into to Time Domain signals

Last day, we analyzed a simple mass spring damper and

found the following: With M = 2, B = 1, and K = 10 in a
coherent set of units, the input force f (t) = u(t), the initial conditions
x (0− ) = 4 and ẋ (0− ) = −1 we had

1 1 8s + 2
X (s) = · + 2
2s2
+ s + 10 s 2s + s + 10
8s2 + 2s + 1
=
s(2s2 + s + 10)

What we’d like to do now is figure out how to use this to find x (t).
We’d like to do so by using the table of common transforms that
we had last day, but this structure doesn’t match any of those, so
we need to make it so that it does. This is where the idea of Partial
Fraction Expansion (PFE) comes in. PFE uses the idea of a common
denominator in reverse. Essentially, if you can take a sum of terms
with different denominators and use cross-multiplication and a com-
mon denominator to add them all up, then you can take a single
term with a complicated denominator and break it up into a sum of
smaller bits.

Example: Consider the following function with independent variable

s. Find A and B so the equation is satisfied. (Hint: simplify the RHS,
then compare like coefficients in the two numerators.)

3s − 10 3s − 10 A B
F (s) = = = +
s2 + 3s + 2 (s + 1)(s + 2) s+1 s+2
136 engg*2400 engineering systems analysis

Great. Now that we have A and B, can you use the RHS of the origi-
nal equation to find f (t)? Hint: use the tables from last day...

You’ve just discovered the principle of PFE and how we use it

to take inverse Laplace :) Is it always this easy? Unfortunately not
(we’ll need to develop special tools for handling complex roots in the
denominator). For now, let’s stick with real, and let’s try something
simple. Recall in unit 2.2 we used calculus to find the solution to a
general first order differential equation τ ẏ(t) + y(t) = kx (t) when the
input was x (t) = x̂u(t) and the initial condition was y(0− ), and we
found that

y(t) = k x̂ (1 − e−t/τ ) + y(0− )e−t/τ , t ≥ 0.

See if you can get the same result using Laplace transform and PFE.
Hint: Start by writing the differential equation in Laplace domain,
then solve for Y (s), the substitute in for your initial condition and
your input. THEN apply PFE and match your terms to the tables. Notice that the tables all have the
leading coefficient in the denominator
as a 1.... for this reason, you’ll make
fewer mistakes when doing PFE and
inverse Laplace if you rearrange your
denominator so that the highest order
coefficient is a one!
 laplace transform 137

Is there a faster way to get these coefficients? Yes! It’s called Heav-
iside coverup and it works in almost all cases.

Example: We want to find the step response for the following second
order, overdamped system:

3s − 10
Y (s) = X ( s ).
(s + 1)(s + 2)

Our first step is to substitute in X (s) = 1/s, then we need to use PFE:

3s − 10
Y (s) =
s(s + 1)(s + 2)
A B C
= + + .
s+1 s+2 s

We can find A, B, and C using the coverup rule. We For example,

3s − 10
A = Y (s) × (s + 1)|s+1=0 = ×
(s 
+ )|s=−1
1
(s 
s  + ) ( s + 2)
1

which is kind of like what we get if we ‘cover up’ the (s + 1) term in

Y (s), then evaluate what’s left behind at the location of that pole. Use
this trick to find the coefficients given above, then find y(t).

So, when can we use this? Any time the roots are real and not
repeated. What happens when they are repeated?

Example: Do PFE on the following

3s − 10
Y (s) =
s ( s + 1)2

First, we need to split it into it’s component parts. Because of the

repeated roots, we need to include a 1/(s + 1) term AND a 1/(s + 1)2
term. We can use coverup for the second one, but not the first... Find
the coefficients:
A B C
Y (s) = + + .
s + 1 ( s + 2)2 s
138 engg*2400 engineering systems analysis

Fantastic! Now let’s go back to that mechanical problem from the

beginning of the lecture....

8s2 + 2s + 1
X (s) = .
s(2s2 + s + 10)

If you check the denominator, you’ll notice that the roots are com-
plex.4 How do we handle this? For now, let’s just factor it out as its 4
This system is underdamped! What do
own term, then see what we can figure out. you expect the response to look like?

8s2 + 2s + 1
X (s) =
s(2s2 + s + 10)
4s2 + 1s + 0.5
=
s(s2 + 0.5s + 5)
A Bs + C
= + 2
s s + 0.5s + 5
A Bs + C
= +
s ( s + a )2 + ω 2
A αω β(s + a)
= + + .
s ( s + a )2 + ω 2 ( s + a )2 + ω 2

Why do I need a Bs and a C term in the

numerator for the second order part?
Okay, now find A, α, β, a, and ω, then use your result to find x (t).
How do B and C relate to α and β?
 laplace transform 139

5.2.1 Assignment:
1. (Drill) From the text, do 7.17, 7.18 (starts pg 237)
2. There is one special case that we didn’t discuss in class: what to do if the order of the numerator is the
same as the denominator. Here, we need a ‘static’ term with no denominator, then we do the rest of
PFE as per usual. To find the static term, just take the highest order coefficient in the numerator and
divide by the highest order coefficient in the denominator, then just do PFE as usual (using coverup if
appropriate). In the following, find the static term A and the other coefficient B, then find f (t). Do so
using two methods: coverup and long division.

3s − 2 B
F (s) = = A+
s + 10 s + 10

3. Let’s go back to system given in the text in problem 12.5 (p 413) on more time. Like before, let the out-
put be the flow wo . In prior assignments you already found the differential equation and the states-
space models of this system. Based on those answers, find the Transfer function, the poles, and the free
response (express your answer in terms of the initial conditions on wo ).
4. You have analysed a hydraulic system and found the following differential equation

ẇo (t) + 4wo (t) = ẇi (t) + wi (t).

Consider the case where the input of interest is wi (t) = 6e−3t u(t) and the initial conditions on the
output are given by wo (0− ) = 9. Using Laplace methods, find the forced, the free, and the complete
responses. Identify the transient and steady state portion of each.
5. From the text, problem 7.7 (pg 241), find the forced and the free responses for the case where the initial
conditions are zero and i (t) = 1, ∀t. Compare this to the answer you got when you found the step
response in way back in section 2.8.5 – they should be different! Why?
6. (Warning, challenging!) From the text, consider the step response given in 8.36. Find the transfer func-
tion of this system
140 engg*2400 engineering systems analysis

5.3 ‘Black box’ modelling

By the end of this section, you will be able to:

• Approximate the system order/damping type based on data from

a step or impulse response

• Use a step or impulse response to find an approximate model of

the system

Most of this course has focused on how to find input

output models if we have lots of information about the
system, but sometimes, we know very little about the ac-
tual system. In fact, sometimes we know nothing at all; essentially,
the system is enclosed in a ‘black box’ and the only thing we can do
is apply inputs, measure the responses, then use the relationships to
find a model. In other words, we have to use what we know about
system responses to reverse engineer the system’s model.
For example, consider the following problem: You want to model
the coupled tanks system from the ENGG*4280 lab, but the flow
is not laminar, so you can’t use the hydraulics element laws from
pre-midterm. You also haven’t taken fluids yet, so you don’t know
Bernoulli’s equation. What you can do is put in some inputs, mea-
sure the outputs, and see what you can figure out based on those
relationships.
The coupled tanks are a complicated interconnected system, so
we’ll look at only one component - the top right tank. What are the
input(s) and the output(s) to that system? (You already did this as a
homework problem!)
 laplace transform 141

Let’s assume that we start with all tanks at steady state and the
voltage of pump 2 set to 6V and pump 1 set to 0V. We then change
pump 2 to 10V and plot the voltage and height on the same graph.

To figure this out, we need to ask ourselves a series of questions: Hint: if we could apply 2400 rules
to this system, what order would we
1. Does this response look (more or less) first order or second order? expect it to be? Maybe that’s a good
place to start?
2. What are the key parameters for that system order?

3. How can I find those parameters from the given input/output

behaviour?

With these questions in mind, find the differential equation that

relates the voltage to the height.

Now ask yourself how you could

validate this result, e.g., what new test
could you run to check if your answer
here was valid?
142 engg*2400 engineering systems analysis

Okay, now that we’ve done a problem with somewhat realistic

data, let’s try a slightly more complicated system, but with perfect
data. Consider the system with the input output behaviour shown
below. Find the differential equation that models the system.

Linear Simulation Results

2
1.8
1.6
1.4
1.2
1
0.8
Input
0.6
0.4
0.2
Amplitude

0
-0.2
Output
-0.4
-0.6
-0.8
-1
-1.2
-1.4
-1.6
-1.8
-2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.51010.51111.51212.51313.51414.515
Time (seconds)

5.3.1 Assignment:
None! Although you might want to think about how you’d solve these problems if you were given a free
response or an impulse response...
 laplace transform 143

5.4 Stability Text: pg 200, 253-257

By the end of this section, you will be able to:

• In your own words, explain the idea of stability using the free
response

• Identify whether or not a system is stable by looking at the pole

locations (for real and for complex poles)

• In your own words, explain why the pole locations determine the
stability of the system

You’ve heard about this idea of stability before, but now

we’re finally going to formally define it: We say that a sys-
tem is stable if its free response goes to zero as time goes to infinity.5 5
Note that this implies that the output
is bounded; i.e., that the output has a
non-infinite maximum and minimum.
Example: Consider the system given by the differential equation
τ ẏ(t) + y(t) = kx (t). Find the free response, then determine the
conditions for which such a system is stable. Assume that k and τ are
any real number (positive or negative!).
144 engg*2400 engineering systems analysis

Example: Find the transfer function for, and determine the stability
of, the system with differential equation

ÿ + 4ẏ + 3y = 2ẋ + x

Hint: Remember that there is an important relationship between the

denominator of H (s) and the denominator of Y f ree ...

Notice that the solution is always a sum of exponents and that

those exponents are always related to the poles of the system... that’s
interesting. Is there a relationship here? Possibly...

Example: Determine the stability of the system with the following

transfer functions (hint: you’ll need to think about how to relate these
to the free response...)

2s + 1
H (s) = .
s ( s + 1)

2s + 1
H (s) = .
(s − 1)(s + 1)
 laplace transform 145

Okay, so we’ve seen some interesting stuff with real poles, what
about complex ones?

2
H (s) = .
s2 + 2s + 2

2
H (s) = .
s2 − 2s + 2

2
H (s) = .
s2 + 2

For all of the above cases, go back and draw the location of the
poles in the complex plane. What can you hypothesize about the
relationship between stability and the pole locations?

We can also add a line to our summary of damping: What can we

say about a system’s stability if ζ < 0?
146 engg*2400 engineering systems analysis

5.4.1 Bounded Input Bounded Output

A neat result of our definition of stability is the follow-

ing: a system is stable iff every bounded input to that system yields
a bounded output.6 In other words, the only way to make the output 6
We call this Bounded Input Bounded
‘blow up’ is to use an input that also ‘blows up’. Output (BIBO) stability.

Earlier, we looked at an un-damped systems and showed that the

free response was bounded, but that it didn’t go to zero, and so it is
unstable. According to BIBO, we should be able to find a bounded
input that yields an unbounded output. Consider the following un-
damped circuit.

iin 8nF 2mH

iL

Based on what you know about resonance and frequency re-

sponse, what bounded input do you think will work? Use Laplace
transform techniques to find the output for your selected input and
show that your output is unbounded. Hint: the transfer function
(measuring time in mirco-seconds) is

1
IL (s) = Iin (s)
16s2 +1
 laplace transform 147

5.4.2 So why does it matter?

Now that we’ve got the math out of the way, based on the
definition of stability and on this neat input/output result, what are
some examples of real systems that are unstable?

In your own words, explain why you think that stability is impor-
tant to a systems engineer:

If you’ve been really observant, you may have noticed a pattern

emerging in our solutions:

1. the free response is always a sum of exponentials where the power

is the poles

2. the forced response is also a sum of exponentials where the power

is the poles plus a term that is related to the input (i.e., if the in-
put is a step, we get a scaled step in the output, if the input is a
sinusoid, we get a sinusoid of the same frequency in the output,
etc.)

Looking at the Laplace transform solutions for the output, can you This is why we care so much about
the system poles, plus it is an extremely
see why this is happening?
powerful sanity check for your solu-
tions!!
148 engg*2400 engineering systems analysis

5.4.3 Assignment:
1. Consider the second order thermal system given in problem 11.9 from the text (p393). Assume that the
ambient temperature θ a is NOT constant and therefore acts as an input, NOT as the reference.
a) Find a state space representation of the system.
b) Find the transfer function of the system.
c) Find the poles of the transfer function and the eigenvalues of the A matrix from your state space
representation. Are they the same or different?
d) Assess the stability of the system. Can you use the state space representation? The Transfer func-
tion? Both? Justify your answer.
2. Are the following systems stable or unstable? If the system is unstable, find a bounded input that yields
an unbounded output. Check your answer by finding the free response and determining whether or
not it reaches a zero at steady state. Hint: You’ll need to convert the transfer function back into the
differential equation, then determine the free response from that...
a)
s+5
H (s) =
(s + 2)(2s + 6)
b)
s−5
H (s) =
(s + 2)(2s + 6)
c)
s+5
H (s) =
(s − 2)(2s + 6)
d)
s−5
H (s) =
( s2 + 5s + 25)
e)
s+5
H (s) =
( s2 − 5s + 25)
Hint: You don’t need to complete the square and do this the ‘long hard’ way... remember that mul-
tiplying by s is the same as taking a derivative... if you know what the free response is for a second
order system, then you can find the free response of s× that system by taking a derivative of the free
response that you already have!
f)
1
H (s) =
s2 + 16
g)
1
H (s) =
s2 − 16
h)
5
H (s) =
s(s + 10)
i)
5s
H (s) =
(s + 10)
 laplace transform 149

j) The system with state space matrices satisfying

" # " #
0 1 8
A= , B=
1 3 12
h i
C= 6 −2 , D = [2].
3. From the text pg. 451, do 13.18c (recall that you already solved a in chapter 3!)
150 engg*2400 engineering systems analysis

5.5 Final Value Theorem and Static Gain Text: pg 233-235

By the end of this section, you will be able to: Secondary Text:

• Apply the Final Value Theorem (FVT) to determine the final values
of time functions

• Identify when the FVT cannot be used

• In your own words, explain the idea of static gain for any stable
system

• Use the FVT to find the static gain of any stable system

There’s one last tool in Laplace that I’ve been hinting

at for a while: the final value theorem (FVT). This relates
back to steady state and to the idea of the static gain. Let’s focus on
steady state first. Notice that, if limt→∞ f (t) exists, then that’s the
same as saying that f (t) reaches a constant steady state. If we have
the function f (t), then it is (usually) easy to find that steady state by
taking the limit, but sometimes we have F (s) and NOT f (t) and we’d
still like to know what the steady state value is. This is where FVT
comes in:
If f (t) has a constant steady state, then

lim f (t) = lim sF (s).

t→∞ s →0

So, if we have F (s) and we have some way of knowing that the
steady state of f (t) is well defined, but we don’t know what f (t)
actually is, then we can use FVT to find it.

Example: Find the steady state of the function f (t) whose Laplace
transform is given by

( s + 2)
F (s) =
(s + 1)(s + 5)

Remember to think about whether or not the steady state exists!

 laplace transform 151

Let’s look at a few more examples to see what we can figure out:

( s + 2)
F (s) =
(s − 1)(s + 5)

( s − 1)
F (s) =
(s − 1)(s + 5)

( s + 2)
F (s) =
( s2 + 1)

( s + 2)
F (s) =
( s2 + 2s + 2)
152 engg*2400 engineering systems analysis

5.5.1 Static Gain

The FVT is particularly useful when we need to know

about the static gain of the system. We’ve talked about the
static gain before, but let’s formally define it now:

Static gain: The static gain of a system is the ratio of the output to
the input once the system has reached a constant steady state.
There are a few key things here, so let’s unpack this definition.
First, the input and output must be constant at steady state (which
is what FVT relates too... cool!). Second, we’re looking for a ratio, so
taking an input of size 1 should be equivalent to taking an input of
size x̂. How does this relate to the differential equation? Consider the
simple first order system:

τ ẏ + y = kx

If the input and output are constant, then what is the derivative of
that output? And therefore what’s the static gain?

Can we extend this to more complicated systems? Sure! Notice

that we already know from our stability discussion that the output of
a stable system looks like a bunch of decaying exponentials plus the
input, so if the input goes to zero or constant, so should the output.
Well that’s nice, but why does it matter? If we think about the forced
response we get the following:

Y f orced (s) = H (s) X (s)

Putting this into the FVT, we get How do we know the FVT can be used
here?
 laplace transform 153

The simplest possible input that has a constant steady state is just
x (t) = x̂u(t), so let’s see what happens when we put that into the Did I need to use x̂, or could I just have
used 1 instead?
FVT:

What does this tell us about the static gain? Given what we already mentioned
about the static gain being related to
when all derivatives are zero, does
it make sense to look at the transfer
function and set all of the s terms to
zero?

Let’s try this on a few systems where we already know the static
gain:
k
H (s) =
τs + 1

kωn2
H (s) =
s2 + 2ζωn s + ωn2

And one more for fun:

3(s + 12)
H (s) =
2s2 + 45s + 6
154 engg*2400 engineering systems analysis

5.5.2 Assignment:
1. From the text, do 8.5 (p288) - focus on part d. Similarly, do 8.19, focusing on c.
2. Determine the final value of output y for the given transfer functions H (s) and input(s) X (s) by using
the FVT. If the FVT does not apply, state why. In all cases, check your answer by using Laplace tech-
niques to determine the response y(t) of the system to the given input.
a)
s 3 5
H (s) = , i ) X (s) = , ii ) X (s) =
s2 + 2s + 1 s s2
b)
s−3 1
H (s) = , X (s) =
s2 + 3s + 2 s
c)
s3 − 8s + 4 1 1 5
H (s) = , i ) X (s) = , ii ) X (s) = + 2
s4 + 19s3 + 191s2 + 1189s + 2920 s s s
d)
s+1 1
H (s) = , X (s) =
s2 − 9s − 10 s
e)
1
H (s) = , X (s) = 1
s2 +9
3. From the text pg. 451, do 13.14b, 13.16d (recall that you already solved a in chapter 3!)
4. Can you use the FVT to find the steady state response if the input is a sinusoid? Why/why not? If not,
what can you use instead?