Download as pdf or txt
Download as pdf or txt
You are on page 1of 125

Mechatronics

Modeling with Bond Graphs


The Practice
of
Multidisciplinary
Systems Engineering

Dr. Kevin Craig


Professor of Mechanical Engineering
Rensselaer Polytechnic Institute
Bond Graphs

K. Craig

Systems and Modeling

What is a system?
What is the systems viewpoint?
What is a subsystem? What is a component?
What is a state-determined system?
What is a model?
Why model?
Analysis?
Identification?
Synthesis?
Hydroelectric Plant

Bond Graphs

K. Craig

IEEE Control Systems Magazine


Bond Graphs

Vol. 14, No. 4, August 1994


K. Craig

Bond Graphs

K. Craig

Engineering System Investigation Process

Engineering
System
Investigation
Process

START HERE

Physical
System
System
Measurement

Parameter
Identification
Physical
Model

Mathematical
Model

The cornerstone of
modern engineering
practice !

Measurement
Analysis

Mathematical
Analysis
Comparison:
Predicted vs.
Measured

Design
Changes

YES

Is The
Comparison
Adequate ?

NO

Bond Graphs

K. Craig

Physical & Mathematical Modeling


Less Real, Less Complex, More Easily Solved

Truth Model

Design Model

More Real, More Complex, Less Easily Solved

Hierarchy Of Models
Always Ask: Why Am I Modeling?
Bond Graphs

K. Craig

Electro-Dynamic Vibration Exciter


Physical System vs. Physical Model

Bond Graphs

K. Craig

Electro-Pneumatic
Transducer

Bond Graphs

K. Craig

This system can be


collapsed into a
simplified
approximate overall
model when
numerical values
are properly
chosen:

Bond Graphs

K. Craig

Temperature Feedback
Control System:
A Larger-Scale
Engineering System
Desired
Temperature
(set with RV)

RV

Block Diagram of an Temperature Control System


eE

Bridge
Circuit

pM

eM
Amplifier

Controller

ElectroPneumatic
Transducer

xV
Valve

TC
Chemical
Process

RC
Actual
Temperature
(measured with
RC)

Bond Graphs

Thermistor

K. Craig

10

Introduction to Bond Graphs


Similar forms of equations are generated by dynamic
systems in a wide variety of domains, e.g., electrical,
mechanical, and fluid. Such systems are analogous.
The bond-graph method is a graphical approach to
modeling in which component energy ports are
connected by bonds that specify the transfer of energy
between system components. Power, the rate of energy
transport between components, is the universal currency
of physical systems.
The graphical nature of bond graphs separates the
system structure from the equations.
Bond Graphs

K. Craig

11

Bond graphs are thus ideal for visualizing the essential


characteristics of a system.
With bond graphs, designing and analyzing the structure
of a system perhaps the most important part of the
modeling task can often be undertaken using only a
pencil and paper. The focus can then be placed on the
relationships among the components and subsystems.
Bond graphs can provide an engineer early with
information about constrained states, algebraic loops,
and the benefits and consequences of potential
approximations and simplifications.
Bond graphs offer qualitative insight to the engineer, in
addition to being used for numerical analysis.
Bond Graphs

K. Craig

12

Collection
of
Engineering
Multiports
Generally, when two
subsystems or
components are joined
together physically, two
complementary variables
are simultaneously
constrained to be equal
for the two subsystems.
Bond Graphs

K. Craig

13

Places at which subsystems can be interconnected are


places at which power can flow between the
subsystems.
Such places are called ports, and physical subsystems
with one or more ports are called multiports.
A system with a single port is called a 1-port, a system
with two ports is called a 2-port, and so on.
The variables that are forced to be identical when two
multiports are connected are called power variables,
because the product of the two variables considered as
functions of time is the instantaneous power flowing
between the two multiports.
Power can flow in either direction, so a sign convention
for the power variables is needed.
Bond Graphs

K. Craig

14

Since power interactions are always present when two


multiports are connected, it is useful to classify the various
power variables in a universal scheme and to describe all
types of multiports in a common language.
All power variables are called either effort or flow.

A curse of system analysis that becomes evident as soon


as problems involving several energy domains are studied
is that it is hard to establish notation that does not conflict
with conventional usage. The context in which the symbols
are used will resolve any possible ambiguities in meaning.
Bond Graphs

K. Craig

15

The power flowing into or out of a port is the product of


an effort and a flow variable.
Momentum is defined as the time integral of an effort.
Displacement is the time integral of a flow.
dp ( t )
P ( t ) = e ( t ) f ( t ) = power
= e(t)
dp = e dt
dt
p ( t ) = e ( t ) dt = momentum
dq ( t )
= f (t)
dq = f dt
q ( t ) = f ( t ) dt = displacement
dt
Energy E(t) is the time integral of power, P(t). p and q
are called energy variables. E t = P t dt = e t f t dt

Effort can be a function of a displacement


Flow can be a function of a momentum
Bond Graphs

() ()
() ()
= e ( t ) dq ( t ) = f ( t ) dp ( t )
E ( q ) = e ( q ) dq
E ( p ) = f ( p ) dp

K. Craig

16

Tetrahedron
of
State
The only types of
variables that are
needed to model
physical systems
are represented by
the power and
energy variables e,
f, p, and q.
Bond Graphs

K. Craig

17

Bond Graphs

K. Craig

18

Bond Graphs

K. Craig

19

e1 = e2 and f1 = f2
power = e1f1 = e2f2
Bond Graphs

Power Bond
The effort and flow signals
are carried by a single
power bond. The half
arrow indicates the
direction of positive power
transport.
K. Craig

20

A word about units:


Units are the shoals on which many a system
analysis has foundered. The International System of
Units (SI) has great advantage for system dynamic
studies.
In the SI system, power is always measured in
newton-meters per second (N-m/s) or the equivalent
watts (W), no matter what type of physical system is
being studied.
Similarly, energy will always be measured in newtonmeters (N-m) or the equivalent joules (J) for any type
of physical system.
Thus if e, f, p, and q variables are given SI units, no
bothersome unit conversions will be necessary to
properly account for power and energy interactions.
Bond Graphs

K. Craig

21

Ports, Bonds, and Power


Multiport elements with ports can be connected to
other multiports to form systems.
When two multiports are connected, power can flow
through connected ports.
We now develop a universal way to represent
multiports and systems of interconnected multiports
based on the variable classifications discussed.
When two multiports are coupled together so that the
effort and flow variables become identical, the two
multiports are said to have a common bond.
A line associated with an isolated multiport indicates a
port or potential bond. For interconnected multiports, a
line represents the conjunction of two ports, i.e., a
bond.
Bond Graphs

K. Craig

22

Separately Excited
DC Motor

Efforts are placed either above or to the left of the port lines.
Flows are placed either below or to the right of port lines.
The half arrow indicates the direction of power flow at any instant when
the effort and flow variable are both positive (or both negative).
Bond Graphs

K. Craig

23

Bond Graphs

K. Craig

24

Bond Graphs
A bond graph simply consists of subsystems linked
together by lines representing power bonds.
When major subsystems are represented by words,
then the graph is called a word bond graph.
Such a bond graph establishes multiport
subsystems, the way in which the subsystems are
bonded together, effort and flow variables at the
ports of the subsystems, and sign conventions for
power exchanges.
Ultimately, detailed bond graphs must be substituted
for the multiports designated by words in a word bond
graph. From a sufficiently detailed bond graph, state
equations may be derived.
Bond Graphs

K. Craig

25

Radar Antenna Pedestal


Drive System

Bond Graphs

K. Craig

26

Automotive
Drive Train

A full arrowhead
indicates an active
bond whose
influence on the
system from its
environment
occurs at
essentially zero
power flow.
Bond Graphs

K. Craig

27

Inputs, Outputs, and Signals


At each port, both an effort and flow variable exist,
and one can control either one but not both of these
variables simultaneously.

Experiment to
determine
what the
motor torque
is at a given
speed.
Bond Graphs

Block Diagram

K. Craig

28

For multiports, each port or bond has both an effort and


a flow, and when these two types of variables are
represented as paired signals, it is only possible for one
of these signals to be an input and the other to be an
output.
To know which of the effort and flow signals at a port is
the input of the multiport, only one piece of information
must be supplied to our previous bond graphs. This is
because if one of the effort and flow variables is an
input, the other is an output.
In bond graphs the way in which inputs and outputs are
specified is by means of the causal stroke.
The causal stroke is a short, perpendicular line made at
one end of a bond or port line and indicates the
direction in which the effort signal is directed.
Bond Graphs

K. Craig

29

The Meaning of Causal Strokes


Note that the half-arrow sign convention for power flow
and the causal stroke are completely independent.
Bond Graphs

K. Craig

30

In many cases, systems are so designed that only


one of the power variables is important, so that a
single signal is transmitted between two subsystems.
No information can really be transmitted at zero
power, but, practically speaking, information can be
transmitted at power levels that are negligible
compared to other system power levels.
Every instrument is designed to extract information
about some system variable without seriously
disturbing the system to which the instrument is
attached.
An ideal ammeter indicates current but introduces
no voltage drop.
An ideal voltmeter reads a voltage while passing
no current.
Bond Graphs

K. Craig

31

+
Vin
-

Bond Graphs

Real
Voltmeter
Input
Impedance

Rin

Ideal
Voltmeter

K. Craig

32

Input
Impedance

Real
Ammeter

Rin
Iin

Bond Graphs

VR

Ideal Ammeter

K. Craig

33

An ideal pressure gage reads pressure with no flow.


An ideal tachometer reads angular speed with no
added torque.
When an instrument reads an effort or flow variable, but
with negligible power, there is a signal connection
between subsystems without the back effect associated
with power interaction.
When a system is dominated by signal interactions due
to the presence of instruments, for example, then either
an effort or flow signal may be suppressed at many
connection points. In such a case, a bond degenerates
to a single signal and may be shown as an active bond.
Notation for an active bond is identical to that for a
signal in a block diagram a full arrow on the bond.
The implication is that the other variable has negligible
backward effect.
Bond Graphs

K. Craig

34

Exercise
Represent an electric drill as a multiport.
Consider the switch position influence as occurring on
an active bond.
Apply causal strokes to your bond graph, assuming
that the drill is plugged into a 100 volt outlet and that
the torque is determined by the material being drilled.
Show a block diagram for the drill corresponding to
your choice of causality at the ports.

Bond Graphs

K. Craig

35

Basic Component Models


We define a basic set of multiports that can be used to
model subsystems in detail.
These multiports function as components of subsystem
and system models and are, in many cases, idealized
mathematical versions of real components.
In other cases, the multiports are used to model physical
effects in a device and cannot be put into a one-to-one
correspondence with physical components of the device.
There are only a few basic types of multiport elements
required to represent models in a variety of energy
domains.
Bond Graphs

K. Craig

36

The bond graph notation often allows one to more easily


visualize aspects of the system than with just state
equations.
Also, the search for a bond graph model of a complex
system frequently increases ones physical
understanding of the system.

Bond Graphs

K. Craig

37

Basic 1-Port Elements


A 1-port element is addressed through a single power
port and at the port a single pair of effort and flow
variables exists.
Here we consider the most primitive 1-ports, elements
that dissipate power, store energy, and supply power.
The 1-port resistor is an element in which effort and flow
variables at the single port are related by a static
function. Resistors dissipate energy. For passive
resistors, establish the power sign convention by means
of a half arrow pointing toward the resistor. Then the
linear resistance parameters will be positive, and the
nonlinear relations will fall in the first and third quadrants
of the e-f plane.
Bond Graphs

K. Craig

38

1-Port Resistor

Bond Graphs

K. Craig

39

Colon Notation
When the component is assumed to be linear, we
indicate this on the bond graph by appending a colon
next to the component, followed by a label.
Also, because the same type of component usually
occurs more than once in a given system, the colon
: notation is used to distinguish between multiple
instances of each component type. The symbol
preceding the colon refers to the component type,
while the symbol following the colon labels the
particular instance.
R:R1 refers to a R component labeled R1.

Bond Graphs

K. Craig

40

1-Port Resistor

Bond Graphs

K. Craig

41

The 1-port capacitor or compliance is an element that


relates e to q. A static constitutive relation exists
between an effort and a displacement. Such a device
stores and gives up energy without loss.
The energy stored in capacitor at any time t is given by:
t

E ( t ) = e ( t ) f ( t ) dt + E 0
0

The energy stored can also be represented as:


q

E ( q ) = e ( q ) dq + E 0
q0

The energy associated with a capacitor is called


potential (or electrical) energy.

Bond Graphs

K. Craig

42

1-Port Capacitor

Bond Graphs

K. Craig

43

1-Port Capacitor

Bond Graphs

K. Craig

44

Stored Energy
for a
1-Port Capacitor

Bond Graphs

K. Craig

45

The 1-port inertia is an element that relates p to f. A


static constitutive relation exists between a momentum
and a flow. Such a device stores and gives up energy
without loss.
The energy stored in an inertia at any time t is given by:
p

E ( p ) = f ( p ) dp + E 0
p0

The energy associated with an inertia is called kinetic (or


magnetic) energy.

Bond Graphs

K. Craig

46

1-Port Inertia

Bond Graphs

K. Craig

47

1-Port Inertia

Bond Graphs

K. Craig

48

Stored Energy
for a
1-Port Inertia

Bond Graphs

K. Craig

49

Tetrahedron
of
State

Bond Graphs

K. Craig

50

The effort source and flow source are simple 1-port


elements. In each case, an effort or flow is either
maintained constant, independent of the power supplied
or absorbed by the source, or constrained to be some
particular function of time.

Bond Graphs

K. Craig

51

A source maintains one of the power variables constant


or a specified function of time no matter how large the
other variable may be and so can supply an indefinitely
large amount of power.
Battery
connected to a
Variable
Resistance
Note: Ideal sources are
useful in modeling real
devices but should not
be expected to be
realistic models in all
power ranges unless
supplemented by other
multiports.
Bond Graphs

K. Craig

52

Rout

Vs

+
-

Bond Graphs

Output
Impedance
Ideal Voltage
Source

+
Vout
-

Real Voltage
Source

K. Craig

53

Iout
Rout
Output
Impedance

Is

Real Current
Source

Ideal Current
Source

Bond Graphs

K. Craig

54

Linear Systems
Bond Graphs

R e = rf
q
C e=
q = f
c
p
I f=
p = e
m

K. Craig

55

Example

Bond Graphs

K. Craig

56

Basic 2-Port Elements


Only two basic types of 2-port elements are required.
There are, of course, an unlimited number of 2-port
subsystems, but we need consider here only those
which cannot be modeled using the basic 1-port
elements and other elements we will define later.
The 2-port elements discussed here are ideal in the
sense that power is conserved at every instant of time.
Whenever power is flowing into one side of a 2-port
element, it is simultaneously flowing out of the other
side.
One 2-port element is a transformer. The constitutive
laws of an ideal 2-port transformer are:
e1 = me 2
mf1 = f 2
m = transformer modulus
e1 ( t ) f1 ( t ) = e 2 ( t ) f 2 ( t )

Bond Graphs

K. Craig

57

Transformers

These are ideal


transformers and in
no case is the
physical device
exactly a transformer.
Bond Graphs

F = AP
AV = Q
A = piston area
K. Craig

58

Actual models of the devices can be made using the


ideal transformer and other multiports to account for nonideal effects if these effects are important to the system
under study.
Another 2-port element is the gyrator. The constitutive
laws of an ideal 2-port gyrator are:
r = gyrator modulus

e1 = rf 2

rf1 = e 2

e1 ( t ) f1 ( t ) = e 2 ( t ) f 2 ( t )

One can show that two gyrators cascaded are equivalent


to a transformer. In contrast, cascaded transformers are
equivalent only to another transformer.
Note that m and r do not have to be constant and so we
have modulated transformers and gyrators. Power is still
always conserved.
Bond Graphs

K. Craig

59

Gyrators

These are ideal gyrators and in no case is


the physical device exactly a gyrator.
Bond Graphs

e = TV
Ti = F
K. Craig

60

Modulated Transformer
Displacement-Modulated
Transformer

y = A sin
y = ( A cos )  or Vy = ( A cos )

( A cos ) F =

Bond Graphs

K. Craig

61

Power Conversion with Transformers and Gyrators


The effort and flow variables within each physical
domain have different units and therefore cannot be
directly connected.
However, since power is the universal currency of
physical systems, the power-converting bond-graph
components TF (generic transformer) and GY (generic
gyrator) provide a means for converting power and thus
connecting different domains.
The TF component generalizes an electrical
transformer, which has the property that the ratio of the
voltages (efforts) at the two terminals is the inverse of
the ratio of current, which is consistent with the fact that
the power is conserved, i.e., instantaneous power at the
input port equals instantaneous power at the output port
at each instant of time.
Bond Graphs

K. Craig

62

e1 = ne2
f1 = nf2

Gearbox
n = gear ratio

Bond Graphs

K. Craig

63

The GY component is the same as the TF component


insofar as power is conserved. The difference is that
flow at one port depends on effort at the other, and
vice versa.
The name gyrator arises from the property of a
gyroscope that angular velocity (flow) is converted into
torque (effort).
In the linear case, the TF and GY components have
the equations:
e2 = kf1
e1 = ne2
GY e = kf
TF f = nf
1
2
1
2
n and k are non-dimensional constants describing the
corresponding physical system.
In both cases, the input and output power is the same.
Bond Graphs

K. Craig

64

e2 = kf1
e1 = kf2
DC Motor
k = back emf constant = torque constant

Bond Graphs

K. Craig

65

3-Port Junction Elements


There are two basic 3-port elements. These are also power
conserving. They allow all energy domains to be assembled
into overall system models. The 3-ports are called junctions,
since they serve to interconnect other multiports into
subsystem or system models. These 3-ports represent one
of the most fundamental ideas behind bond graph formalism.
The idea is to represent in multiport form the two types of
connections called series and parallel connections.
The first is the flow junction, 0-junction, or common effort
junction. The efforts on all bonds of a 0-junction are always
identical and the algebraic sum of the flows always vanishes.
Taken together, the equations imply that power on all the
bonds sums to zero. Power is neither dissipated nor stored.
Bond Graphs

K. Craig

66

The second is the effort junction, 1-junction, or common


flow junction. The 1-junction has a single flow, and the
sum of the effort variables on the bonds vanishes. As
with the 0-junction, the power on all bonds sum to zero.

Physical
Interpretation
of
0- and 1-junctions
in several domains

Bond Graphs

K. Craig

67

Basic 3-Ports

0-Junction

1-Junction
Bond Graphs

K. Craig

68

Summary of Basic 3-Ports

Bond Graphs

K. Craig

69

Examples of Systems
Involving
Basic 3-Ports
Note:
A 0-junction
has only a
single effort
and a 1junction has
only a single
flow,
independent of
the sign
convention.
Bond Graphs

K. Craig

70

0
common-effort
junction

e1 = e2 = e3
f1 f2 f3 = 0
Bond Graphs

1
common-flow
junction

f1 = f2 = f3
e1 e2 e3 = 0
K. Craig

71

Mechanical
Series & Parallel
Components

x 1, v 1
K

f, v

Series Connection

f, v

Parallel Connection

When two elements carry the same force they are


said to be connected in parallel. When two
elements have the same velocity they are said to be
in series.
Bond Graphs

K. Craig

72

Summary
The symbols 0 and 1 are chosen to be neutral with
respect to the physical domain.

Bond Graphs

K. Craig

73

Example: Simple RCI System

Bond Graphs

K. Craig

74

Bond Graphs

K. Craig

75

Fluid Resistor
Pure and Ideal Fluid Resistor

q=

p 2 p1
Rf

Most fluid resistors are nonlinear.


A sharp-edged orifice is an example.

2
p 2 p1 =
q
2
2A1
A2 >> A1

Bond Graphs

K. Craig

76

A pure and ideal fluid resistor behaves exactly like


a pure and ideal electrical resistor when the
voltage-pressure and current-flow analogy is used.
Fluid Capacitor
Fluid capacitors are found in numerous hydraulic
and pneumatic systems. Examples include
reservoirs, pressurized tanks, spring-loaded
accumulators, and air-charged accumulators.
An open reservoir is often used in a hydraulic
system as a capacitor.
A spring-loaded accumulator is another example.

Bond Graphs

K. Craig

77

dh
Conservation of Mass
dt
ghA
p=
pressure at the tank bottom
= gh
A

q=A

dh 1 dp
=
dt g dt
A dp
=
q
g dt
dp
q = Cf
dt

Bond Graphs

K. Craig

78

In this spring-loaded
accumulator, a spring rather
than gravity provides the
pressure increase. A volume
flow rate entering the bottom of
the tank causes the spring to
compress a distance x. This
increases the pressure p in the
tank.

dx
dt
kx
p=
A

q=A

Conservation of Mass
pressure in the tank

dx A dp
=
dt k dt
A 2 dp
q=
k dt
dp
q = Cf
dt

Bond Graphs

K. Craig

79

The energy stored in a fluid capacitor is given by:

1
E = Cf p 2
2
Fluid Inductor
A mass of fluid is quite similar to a solid mass in
motion. The fluid mass has inertia and a force is
required to accelerate or decelerate the fluid.
Consider an ideal (no viscosity
and hence no friction forces)
incompressible fluid in an
unsteady (flow velocity is not a
constant) flow through a pipe.

Bond Graphs

K. Craig

80

The fluid mass is A, the net force acting on the


mass is A(p2 p1), and the acceleration of the fluid
mass is dU/dt.
dU
nd
From Newtons 2 Law
A ( p 2 p1 ) = AA
dt
dU A dq
p 2 p1 = A
=
dt A dt
dq
p 2 p1 = Lf
dt
A pure and ideal fluid inductor behaves exactly
like a pure and ideal electrical inductor.
The energy stored in a fluid inductor is given by:
1
E = Lf q 2
2
Bond Graphs

K. Craig

81

Bond Graphs

K. Craig

82

Causality Considerations for the Basic Multiports


Some of the basic multiports are heavily constrained with
respect to possible causalities, some are relatively
indifferent to causality, and some exhibit their constitutive
laws in quite different forms for different causalities.
Causality for Basic 1-Ports
A source impresses either an effort or flow time history
upon whatever system is connected to it.
In the linear case, with a finite slope of the e-f
characteristic, the 1-port resistor is indifferent to the
causality imposed on it.
In considering the causality of the 1-port capacitance and
the 1-port inertia, integral causality and derivative causality
result.
Bond Graphs

K. Craig

83

Causal Forms for Basic 1-Ports

Bond Graphs

K. Craig

84

The causal stroke


perpendicular end bar
indicates the direction in
which the effort signal is
directed.
The bond-graph
representation is acausal and
represents an equation. The
addition of a causal stroke
assigns the input and output
of each R component. This
causal assignment is not part
of the initial modeling but is
added later. Bond-graph
components are reusable
within different causal
contexts, whereas block
diagram components are not.
Bond Graphs

K. Craig

85

Integral causality

Derivative causality

Bond Graphs

K. Craig

86

Derivative causality

Integral causality

Bond Graphs

K. Craig

87

Causality for Basic 2-Ports and 3-Ports


For a transformer, there are only two possible causality
assignments . As soon as an e or f has been assigned as
an input to the transformer, the other e or f is constrained to
be an output.
Similarly for the gyrator, as soon as the causality for one
bond has been determined, that for the other is also.
The causal properties of the 3-port 0-junction and 1-junction
are somewhat similar to those of the basic 2-ports.
Although each bond of the 3-ports, considered alone, could
have either of two possible causalities assigned, not all
combinations of bond causalities are permitted by the
constitutive relations of the element.
Only when some real system models have been assembled
is it clear why causal information is so important.
Bond Graphs

K. Craig

88

Causal Forms for Basic 2-Ports and 3-Ports

Bond Graphs

K. Craig

89

For a 3-port 0-junction there are only three different


permissible candidates in which each of the three bonds
in succession plays the role assigned to bond 1 in the
previous table. For a n-port 0-junction this description of
the constraints on causality is still valid and there are
exactly n different permissible causal assignments.
For a 3-port 1-junction the same considerations apply as
for a 3-port 0-junction except that the roles of the efforts
and flows are interchanged. Clearly, there are three
permissible causalities for a 3-port 1-junction, and there
are n permissible different causal assignments for a nport 1-junction.

Bond Graphs

K. Craig

90

Causality and Block Diagrams


Block diagrams indicate input and output quantities for
each block and thus are inherently causal.
When causal strokes are added to a bond graph, one
may represent the information by a block diagram.
It should be possible to correlate the signal flow paths in
the block diagrams with the equations in the tables and
with the bond graph representation.
Note that when one rigorously maintains the spatial
arrangements with efforts above and to the left of bonds
and flows below and to the right, the block diagrams
have fixed patterns.
Block diagrams are more complex graphically than bond
graphs because a single bond implies two signal flows.
Bond Graphs

K. Craig

91

Block diagrams have an unfortunate drawback; they


represent assignment statements rather than equations.
A block diagram cannot be drawn until the inputs and
outputs of each component are specified.

Bond Graphs

K. Craig

92

Block Diagrams
for
1-Ports

Bond Graphs

K. Craig

93

Block Diagrams
for
2-Ports

Bond Graphs

K. Craig

94

Block Diagrams
for
3-Ports

Bond Graphs

K. Craig

95

System Causality

RCI System

Bond Graphs

K. Craig

96

The 0 junction corresponds to the 1st summation block.


The 1 junction corresponds to the 2nd summation block.
Bond Graphs

K. Craig

97

Interconnected
Drive Train
Model:
Bond Graph
and
Block Diagram

Bond Graphs

K. Craig

98

Exercises
Draw block diagrams for the following bond graphs ,
assuming that all 1-ports are linear.

Bond Graphs

K. Craig

99

System Models
We are now ready to model the world armed with our
bond-graph arsenal of C, I, R, Se, Sf, 0, 1, TF, and GY.
However, it is not true that every system you may
encounter will be reducible to a simple bond graph. But
the number of physical systems which can be
represented by a bond graph is amazingly very large.
We initially will look at single-energy-domain systems,
i.e., systems that involve only one type of power, e.g.,
electrical, mechanical, hydraulic.
Devices involving two or more types of power have
transducer elements coupling the different energy
domains.
Bond Graphs

K. Craig

100

Electrical Systems
Any electrical circuit can be modeled by a bond graph
containing elements of the set {0, 1, C, I, R, Se, Sf}.
Notice that the elements TF and GY are not included. That
is because these elements are properly used in
representing electrical networks, a more general class than
circuits.
First we will model circuits, then networks.
So the question is how do we use the junction elements to
construct an overall bond graph model of an electrical
circuit. Sometimes for simple circuits it is easy to
recognize that some elements have the same current
(flow) and others have the same voltage (effort). For these
circuits, bond graph construction can be accomplished by
inspection.
Bond Graphs

K. Craig

101

Simple Electric Circuit

Bond Graphs

K. Craig

102

To arrive at the bond graph, we proceed as follows:


C and R1 have the same voltage and thus are attached to
the same 0-junction.
L and R2 have the same current and so are attached to
the same 1-junction.
The bond joining the 0-junction and 1-junction enforces
the fact that the current through the inductor is the sum of
the currents through the capacitor and resistor R1.
All the 1-port R, C, and I elements have the power halfarrows defined such that whenever the voltage drop
across the element is in the direction defined as positive
and the current is simultaneously in the defined positive
direction, the power is flowing into the element. We
always define positive power directions for R, C, and I
such that this is true.
Bond Graphs

K. Craig

103

Most of the time, electric circuits are too complex to model


by inspection. There may be parts that are obviously in
series or parallel (common current or common voltage), but
constructing the overall bond graph model is much easier if
a procedure can be followed that ensures success
regardless of the complexity of the circuit.
So lets develop a fool-proof circuit construction procedure
with the aid of an example.
Step 1: Assign a power convention to the circuit schematic.
This step must always be done. Show the positive
voltage drop and current directions. For the I, R, and C
elements, the positive voltage drop is shown in the
same direction as the positive current. This ensures
that power is directed inward on the corresponding bond
graph element.
Bond Graphs

K. Craig

104

For the source elements, it is not critical which


directions are chosen for positive voltage drop and
positive current. If positive current is defined such that
the current moves uphill against positive voltage,
then positive power will come from the source into the
rest of the circuit. If either the positive voltage
direction or current direction is chosen in the opposite
direction, then positive power will be absorbed by the
source.
Step 2: Label each node voltage on the circuit schematic
and use a 0-junction to represent each node voltage.
A node voltage is the voltage above and below or the
left and right of each circuit element. Label the node
voltages using letters. Every bond that touches a
particular 0-junction has an identical voltage.
Bond Graphs

K. Craig

105

Step 3: Establish the positive voltage drops across the


elements using the 1-junctions.
Remember that 1-junctions add efforts (voltages)
according to the power convention. By properly
directing the half arrows on 1-junctions, the proper
voltage drop can be established across each bond
graph element.
Step 4: Remove all bonds that have zero power.
Before the bond graph can be used for equation
derivation or simulation, the reference voltage must be
established. Our reference is ee and it is zero since it is
the ground voltage. Since every bond that touches a 0junction has the identical voltage, all bonds inside the
curve on the figure have zero voltage and each of those
bonds carries no power.
Bond Graphs

K. Craig

106

We can either append an effort source of zero voltage


to one of the 0-junctions representing ee, or we can
simply erase all the bonds that carry no power.
Step 5: Simplify the bond graph by using the bond graph
identities.
This is not an absolutely necessary step. By
removing the 0-junctions and 1-junctions with a
through power convention, a much neater picture
emerges.
Also the loop structure eb to ec has been reduced.
We establish the voltage drop eb-ec once, and then
attach I and C elements associated with L1 and C2
to a 0-junction constrained to have this voltage drop.

Bond Graphs

K. Craig

107

Electric Circuit Example

Step 1

Step 2

Bond Graphs

K. Craig

108

Step 3

Step 4

Bond Graphs

K. Craig

109

Step 5

Bond Graphs

K. Craig

110

Exercise
Consider the Wheatstone bridge. This circuit is
typically used with strain gages as the resistive
elements, R1 through R4, and the voltage across the
load resistance, RL, is the output that is indicative of
any change in the bridge resistances.
Step 1 is completed. Construct the bond graph.

Bond Graphs

K. Craig

111

Bond Graphs

K. Craig

112

Electrical Networks
An electrical network is an extension of electrical circuits
to include transformers and gyrators.
An electrical transformer is a common electromagnetic
device used to step voltages up or down while doing the
opposite to current.
Electrical gyrators are exhibited in Hall effect transducers
where voltage across a semiconductor material is related
to a current through the material perpendicular to the
voltage drop direction.
The basic rules for bond graph construction remain
unchanged.

Bond Graphs

K. Craig

113

The figure shows the electrical symbol for a transducer


where N indicates the turns ratio across the device.
Positive voltage drops and current directions are chosen
such that positive power is into the device on the left side
and out of the device on the right side.

Bond Graphs

K. Craig

114

An electrical network with an isolating transformer is


shown below. Positive voltage drops and current
directions are shown. Notice that positive power flows in
on the left side of the transformer and out on the right
side.

Bond Graphs

K. Craig

115

Bond Graphs

K. Craig

116

Bond Graphs

K. Craig

117

State-Space Equations
Bond graphs are an acausal representation. By
assigning a causal stroke to each bond, a causal
representation can be generated.
The causally complete model can be converted into
other causal representations such as state-space
equations and block diagrams.
The causal strokes on a bond graph provide sign posts
to guide the generation of state-space equations and
block diagrams. State-space equations can be
generated by hand.

Bond Graphs

K. Craig

118

Causal Assignment
Abstracting the physical system as an acausal bond
graph provides a complete description of the
corresponding model.
There are many ways of representing the system as a
set of equations. The state-space representation for
system analysis and simulation is preferred.
For proper causal completion, which will result in a
set of explicit assignment statements, it is necessary
that exactly one bond impose a flow on each 1
junction.
Similarly, exactly one bond must impose an effort on
each 0 junction.

Bond Graphs

K. Craig

119

The causality of TF and GY components is also subject


to constraints if self-consistent models are to be
generated. In particular, causality is transmitted
unaltered through TF components, that is, one
impinging bond imposes effort (flow), while the other
has effort (flow) imposed on it.
Causality is reversed through a GY component so that
both impinging bonds impose effort (or flow) and have
flow (or effort) imposed on them.
Within these guidelines, causality can be assigned
arbitrarily.
After specifying the causality at the external interfaces, it
is generally advisable for the modeler to specify the
preferred causality of the system C and I components,
which may have integral or derivative causality.
Bond Graphs

K. Craig

120

For simulation or state-space representations,


integral causality is usually preferable since it
leads to ordinary differential equations.
Bond graphs can help engineers decide which
approximations are useful before generating the
equations, e.g., deciding whether to model a shaft
connecting two rotating masses as either rigid or
compliant.

Bond Graphs

K. Craig

121

Steps:
Identify the states
The system states are the
integrated flows q associated
with C components and the
integrated efforts p associated
with the I components, in integral
causality. Here, they are q1 and
p1.

Write state derivatives in terms


of states and inputs
dq1
= fc
dt

p1
f1 =
m1

f c = f o f1

dp1
= em
dt

e m = eo e r e1

eo =

q1
p
e r = r1f1 = r1 1
c1
m1

Write outputs in terms of states


p1
and inputs
=
f1
Bond Graphs
m1

dq1
p1
= f0
dt
m1
dp1 q1
p1
= r1
e1
dt c1
m1
e0 =

q1
c1

K. Craig

122

Bond Graph Case Study

Bond Graphs

K. Craig

123

Bond Graphs

K. Craig

124

Bond Graphs

K. Craig

125

You might also like