You are on page 1of 33

Chapter 1

Introduction to Bond graphs


1.1 Introduction

Unlike his predecessors a few generations ago, modern man has to deal with variety,
diversity and is compelled to be increasingly innovative. Need for growth has
become essential for survival. Though it may sound paradoxical, it is true that the
only way to deal with variety and diversity and to sustain the flow of innovation is to
achieve conceptual unification. Again conceptual unification is often made to shrink
to techniques of creating analogues of unmanageable or unfathomable domains into
fields with established empirical structures. Such an art of analogue making is not
unification. Unification, in fact, should imply uniformity of symbolism, dialectics and
deductive processes dissolving the partitions of experience. The requirements of a
unified approach to modeling, simulation and synthesis of physical systems could be
stated as follows.

(a) It should have complete but concise lexicon with validity over a large variety
of diverse domains.
(b) Allow the modeler to portray the interactions within each and exchange across
these domains, using aforesaid lexicon.
(c) The portraits thus created should algorithmically lead to mathematical or
logical models, which may be at a higher level of abstraction (i.e., without
needing any intuitive trickery). These models may then be subjected to
predictive or deductive processes.

In physical systems it is energy which plays the role of common currency of


exchange between various domains and sustains the business of dynamics. It is
surprising that Newton's most prominent contemporary, Leibnitz [71] realized this
fact. The statement of Leibnitz may appear rather strange to modernist; still the
underlying idea would not be missed even at a casual glance. The statement is “The
forces are of two kinds namely dead and live. The dead force depends on position and/or on
configuration and the live force is proportional to square of velocity. The sum of these two
forces in the universe remains constant”.

Immediate successors of Newton and Leibnitz (John Bernoulli, Daniel Bernoulli et al)
found that many problems of mechanics are formulated and solved with greater ease
by this idea of Leibnitz than by the methods of Newton. If we replace the term force
by energy the statement of Leibnitz is law of conservation of energy in mechanical
systems.
INTRODUCTION TO BOND GRAPHS 2

The person who firmly founded mechanics on the concept of energy was J. L.
Langrange. In 1788 he published a book entitled "Mechanique Analytique" [38]. To
emphasize the liberation from tentative trickeries, the role of intuition and
geometrical constructions, Langrange did not use any drawing or diagram in his
book. What Langrange achieved was a perfect algorithmic (which he chose to call
analytical) approach to mechanics. However, in 1788 unification was not in sight. No
other domain of energy was identified as being dynamic. Heat was still juicy
phlogiston. Though Charles Du Fay and Benjamin Franklin had classified static
electricity, the influence of motion of charge was unknown. One may say with
Faraday, Ampere, Ohm and then Joule, Fourier et al the boundaries of classical
dynamics started expanding. Need for unification became stronger than ever before.

Around 1960 two remarkable developments took place in the field of classical system
modeling. T.R.Kane [30] invented ways to overcome the serious limitations of D'
Alembert - Lagrange mechanics. Kane's method however remains algebraic like
those of both D' Alembert and Lagrange not so efficient for deployment of modern
computers at the initial stages of model making. During the same period, H. M.
Paynter [47] invented bond graphs. The method of Paynter or bond graphs render all
that has been stated as basic requirements of a modeling technique, it is both
unifying and algorithmic.

1.2 An invariant nature of power


exchange

The exchange of power between two parts


of a system has an invariant characteristic.
To understand this invariant nature, let a
simple system made of two elements be
considered. Let the system belong to the
realm of mechanics, involving forces,
velocities etc. The elements of the system
are, say, A and B as shown in Fig. 1.1.
Fig. 1.1

At a particular instant, assume that A is driving B i.e., A is imparting power to B.


This is shown in Fig. 1.1 by a half-arrow. If A had some consciousness, its observation
would be that it is pushing B towards EAST by applying a force in that direction and
B is moving towards EAST. On the other hand, if B had a similar consciousness, its
observation would have been that it (B) is applying a push towards WEST on A but it
has to retreat towards EAST (the loser is gaining energy!). So, the fact that power is
going from A to B is viewed by A and B in two different ways, as regards the
directions of force and velocity. Now we state the basic idea - the direction of power
3 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

flow at any moment is a system invariant. The force (effort) and the velocity (flow)
are its factors. Their relative disposition depends on the stand point the analyst takes
i.e., whether he assumes himself to be A or B.

In addition to the above, energy, the accumulation of power, has an interesting


property that it is set-additive or a measure (in mathematical sense) or an extensive
quantity. The precise meaning is that the total energy of a system is sum of the
energies stored in its exclusive components. The time rate of energy, i.e., power, then
has a derived property that the total power entering in a system at any point of time
is sum of power reaching its components which store it and power rejected through
the dissipaters.

A system can, nearly always, be conceptually dichotomed in two sets, one which
consists of energy storers, dissipaters and importers and the other which signifies the
constraint structure. This constraint structure neither creates nor destroys power.
Paynter created a set of powerful symbolisms which express this deep, though
abstract interplay of power and energy in a concrete and discernible form, through
diagrams which represent the essential dynamics of the systems and aid creation of
system equations and prediction.

Many books and articles which introduce bond graphs often tend to present them as
something which is equivalent to other system representation techniques like block
diagrams and signal flow graphs. Bond graphs are, however, independent
formulation of classical system dynamics which reveal many obscured structures in
it, unify it and most significantly stimulate the imagination of the modeler by
manifestation of concrete and abstract ideas in integrated forms. Other
representations of a system, however, may be derived from a bond graph. Bond
graphs are far more powerful in modeling complex systems even with interaction of
several energy domains.

There are several organized books on bond graphs and their applications viz. [10, 21,
31, 32, 53, 68] with varied emphasis and style. In this book we attempted to present
both, the abstract ideas on which this powerful modeling technique is founded and
its applications in various domains of energy and their interactions.

1.3 Bond graphs - a first impression

Alvin Toffler in his article "Science and Change" which is a preface to the celebrated
book "Order out of chaos" by Ilya Prigogine and Isabelle Stengers [50] says:-

“One of the most highly developed skills in contemporary western civilization is dissection:
the split-up of problems into their smallest possible components. We are good at it. So good,
we often forget to put the pieces back together again. This skill is perhaps most finely honed in
INTRODUCTION TO BOND GRAPHS 4
science. There we not only routinely break problems down into bite-sized chunks and mini-
chunks, we then very often isolate each one from its environment by means of useful trick...”

As science has now occupied the central arena of world culture, dissection has
become pervading world tendency, where as system synthesis is predominantly the
art of putting the pieces together. What is needed is a technique which works both
ways – bond graph does that.

A modeler breaks a system and ascertains the basic entities or dynamical units it is
composed of, namely the inertances, compliances and dissipaters or resistors and
also sources external to the system. An option of refinement remains in terms of
more number of inertances, compliances and dissipaters, albeit, at one stage further
refinement is stopped, either by satisfaction from the extent of incorporated details
or due to the limitations of handling resources. These elements become his atoms.
Now comes the stage of putting the pieces together. This is done by portraying the
constraint structure binding these atomic elements to get back the system.

In bond graph theory the basic elements i.e., inertances, compliances (capacitance)
and dissipaters are represented by I, C and R respectively. External sources of input
are represented by SE and SF for generalized efforts and flows respectively. The
constraints are represented by using an additional set of elements. With these, a
portrayal of the system is created that represents the constraint structure and binding
of the atomic elements and sources. The elements of both these classes are connected
by line segments called bonds. The bonds portray the paths of exchange of power
within the constraints structure and to the atomic elements. The diagram thus
evolved is called a bond graph of the system.

The above description is only intended to give a first impression of the bond graphs.
A deeper insight will grow as we go along considering in detail various aspects of
this system modeling tool and its application.

1.4 Power variables of bond graph

It was indicated earlier that power may be expressed as multiplication of two factors.
One of these factors was classified generalized effort ( e ) and other as generalized
flow ( f ). In multidimensional cases it is the inner product, scalar product or dot
product of effort and flow vectors which gives the power.

Let us view the matter in a different manner. The change in energy is integration of
power.
t t
E   1 P ( )d   1 e( ) f ( )d , (1.1)
t0 t0
5 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

where P ( ) is power.
Two interesting alternative ways in which the above integration may be written are
as follows

t2 Q (t 2 )
dQ
E   e( ) d   e(Q )dQ, (1.2)
t1 d Q (t1 )

t2
dP p (t 2 )
or E   t1 d
f ( ) d   p ( t1 )
f ( p ) dp. (1.3)

In the first expression (1.2) the factor f is assumed to be rate of change of generalized
displacement Q and the variable of integration is changed to generalized
displacement. Factor e, which is now the integrant, is assumed to be function of this
displacement.

The usual or a direct approach to classical dynamics emerges from this view point.
The generalized displacement and its time rate of change, taken as generalized flow,
are the basic dynamical variables. The generalized effort which is its companion
factor in power expression is related to these variables.

In direct formulation, the energy of a linear compliant element becomes

Q1
1
E   KQ dQ  ( KQ12  KQ22 ). (1.4)
Q0 2

The energy stored in an inertial element of mass m may be obtained as follows

t1 t1
E    dt 
eQ   Q
mQ  dt
t0 t0
(1.5)
 
 (t )
Q
  1 mQ
1
  dQ
mQ  2 (t )  m Q
 2 (t )
1 0

Q (t0 ) 2

Usually such an expression for energy stored in an inertial element is called kinetic
energy. Crandall et al [12] preferred to call it co-kinetic energy.

The following table gives the effort and flow variables for the direct formulation in
some physical domains. In this book we will mainly confine ourselves to direct
formulation of classical system dynamics.
INTRODUCTION TO BOND GRAPHS 6

Table 1.1

Power variables in some energy domains.


Systems Effort(e) Flow(f)
Mechanical Force(F) Velocity(v)
Torque(τ) Angular velocity (  )
Electrical Voltage(V) Current(i)
Hydraulic Pressure(P) Volume flow rate( Q  )

Thermal Temperature(T) Entropy change rate( S )


Pressure(P) Volume change rate(  )
Chemical Chemical Potential() Mass flow rate ( m
 )
Enthalpy(h) Mass flow rate( m  )
Magnetic Magnetic -motive force(em) 
Magnetic flux rate (  )

The integration (1.3) gives rise to an alternative formulation of classical system


dynamics called the dual formulation. The momentum p is given the status of
generalized displacement. Usual effort, p is then treated as flow and its companion
in power expression (the usual flow) is treated as effort. Such a formulation was
proposed by Trent and Toupin [70, 64]. For some interesting applications of this
formulation and theoretical aspects associated with it readers may refer to [54, 55].

The energy of a mass point in this formulation would be

dp p 1
 
t1 p ( t1 )
E   f dt   dp  p 2 (t1 )  p 2 (t 0 ) .
t0 dt p ( t0 ) m 2m
(1.6)

Crandall et al [12] called this as kinetic energy of the system. For a linear inertial
element both these expressions render the same value.

The energy stored in a linear compliant element in such a formulation may be


obtained as follows.
7 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

 
t1 t1 e ( t1 )
de e 1
E   e f dt   e dt   de  e 2 (t1 )  e 2 (t0 ) .
t0 t0 Kdt e (t 0 ) K 2 K
(1.7)

With the generalized momentum acquiring status of generalized displacement and


generalized effort acquiring that of generalized flow, the expression of stored energy
in an inertial element appears like usual stored potential energy and that in a
compliant element like usual kinetic energy of a mass point. Infact in dual
formulation compliant element behave like inertial elements of direct formulation
and vice-versa.
We would like to remind the readers that this book is entirely based on direct
formulation of classical system dynamics.

1.5 Bond graphs for simple electrical circuits

To make a bond graph, one has to keep track of power. Let us consider a simple
circuit with a resistor, a capacitor, an inductor and a source of voltage, all connected
in series as shown in Fig. 1.2(a). Could a modeler have refined this system further?
Well, yes. He could have considered the capacitor to be a leaky one by putting a high
resistor in parallel with it, or by considering parasitic capacitors at each of the tag
points of the circuit or between the coils of the inductor and so on. But our modeler
chose to stop at the level shown in the figure. The justification being that all these
aspects would show a discernible influence either over a very long duration of
operation of the circuit or at very high frequencies. Both these ranges do not interest
the modeler.

Fig. 1.2(a) Fig. 1.2(b)

The resistor throws power to the thermal domain. The path this power takes from the
system to the environment cannot be determined. Even the mechanism of this
conversion is extremely complex and calls for good deal of quantum solid state
physics. However, we impose an abstract line signifying the passage of power to an
element R as shown in Fig. 1.2(b). R iconizes the action of the resistor.
INTRODUCTION TO BOND GRAPHS 8
Power gets stored in the inductor. The flow of current through the inductor gives rise
to magnetic field. The current flowing in that magnetic field leads to this storage of
energy. The phenomenon of such a storage is perfectly reversible (as no heat is
produced) and the energy may revert back to the circuit. We draw an abstract line
leading to an element I, where I, likewise, iconizes the action of the inductor.

The capacitor too stores energy. The displacement of charge on its plates gives rise to
electrical field. The charge residing in this field causes storage of energy. This
phenomenon is also reversible. An abstract line leading to an element C, represents
this storage where C iconizes the action of the capacitor. The transmission of power
from the source is represented by a line connected to an element SE. SE represents a
source of generalized effort, in this particular situation a source of potential
difference. We again emphasize that the lines leading to the various elements are
abstract representations of paths the power takes and are called bonds. The factual
path taken may be quite different. Say for example, if we consider power going to a
capacitor, the path taken by it is fairly counter intuitive.

Fig. 1.2(c) Fig. 1.2(d)

Figure 1.2(c) shows a capacitor being charged. The power follows in the direction of
the Pointing vector E  B which is radically inwards. Now we refer to Fig. 1.2(b)
and remove the circuit and retain the power paths or bonds as shown in Fig. 1.2(d).
Let us also number the bonds for referring them. Each of these bonds is associated
with two power variables, generalized effort e and generalized flow f. The constraint
imposed by the circuit is such that the flows (currents) are equalized in the bonds i.e.,

f1 = f2 = f3 = f4
The constraint of equalization of the flows is represented by inserting an icon S at the
junction point as shown in Fig. 1.2(d). S (called S-Junction) is nothing material; rather,
it is an abstract representation of the constraint of equality of flows along the bonds
connected to it. The equality, we emphasize, is quantitative and not qualitative. This
9 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

fact is of extreme significance for unified system modeling. We elaborate this thought
as follows. The current flowing through resistor is current carried by charges or
electronic current. Same is true for the inductor. But, what does flow through the
capacitor, which is a discontinuity in the flow path of the charges?

A great genius of science, J. C. Maxwell was possessed by this question and came up
with an answer which not only enlarged the horizon of our understanding of the
physical world but showed a path to the later generations towards unification of
world picture. He enlarged the model of electrical field

B
   E  ,   B  0,  E   and   B    
t
B
to    E   ,   B  0,  E   and
t


  B     
E
t



Displacement current

The usual electronic


current
where

E= electric field vector  = permittivity


B= magnetic induction = magnetic permeability
v = Velocity of the charge = charge density
He appended an additional current, non-electronic in nature called displacement
current. Displacement current is proportional to rate of change of electric field. So
what flows through the capacitor is displacement current and thus what S signifies is
quantitative equality and not qualitative equality.

Let us consider a parallel R-L-C circuit with a modification. The source is now
replaced by a source of current. Current being flow in the direct formulation, this is
represented by a source of flow SF. The circuit and its bond graph model are shown
in Fig. 1.3 (a) and Fig.1.3 (b).
INTRODUCTION TO BOND GRAPHS 10

Fig. 1.3(a) Fig. 1.3(b)

The new feature of this bond graph is the P-junction to which the elements are
attached. The junction P represents the constraint of equal potential difference across
the elements.

If S is said to represent equal flow junction, P stands for equal effort junction in
general bond graph models.

A somewhat complex circuit and its bond graph model are shown in Fig.1.4 (a) and
Fig. 1.4 (b).

Fig.1.4(a) Fig. 1.4(b)

Another such circuit and its bond graph model are shown in Fig.1.5 (a) and Fig.1.5(b).

Fig.1.5 (a) Fig.1.5 (b)


11 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

We encourage the reader to discover the logic which has led to the creation of these
bond graphs. However, in Chapter - 4, step by step procedures to create system bond
graph models are discussed. We recommend the reader to go through that chapter
once he is introduced to the notion of power directions.

Reader may notice that bonds are numbered in the model. This is essential to carry
on further analysis of bond graph models and to ‘book-keep’ them. The elements of
the model now attain two types of identification one physical and the other bond
graph generic. The physical identification of an element relates to the name of the
element used in the physical model of the system. This name is indicated in the bond
graph model by inserting a colon (:) near the element in the bond graph. The generic
identification is derived from the bond number to which the element is appended.
Thus, the physical identification of the resistive element on bond number 2 in Fig.
1.4(b) is R1 where as its bond graph generic identification is R2. In the analysis of a
given bond graph model it is advantageous to use the bond graph generic names.
Once the final mathematical models are arrived at in terms of generic names these
may be replaced by their physical name. The software analyzing bond graph
manages this by simply renaming the parameters to be given by the modeler.

1.6 Representation of junction elements

The originators of bond graphs did not wish that the physical topology of electrical
series circuit should become conceptual intaglio for common flow junction and
likewise that of parallel circuit for common effort junctions. They advocated use of
neutral symbols which are related to nothing physical. 1 is used for common flow
junctions and 0 for common effort junctions. The bond graph of the system shown in
Fig.1.6 (a) would then be as shown in Fig. 1.6 (b) and that of the system shown in Fig.
1.7 (a) be as shown in Fig. 1.7(b).

Fig.1.6(a) Fig.1.6(b)

Authors of this book often argued that 1 and 0 were not the best selections. Numerals
are often used either in cardinal or ordinal sense, signifying population or order. The
INTRODUCTION TO BOND GRAPHS 12
junctions do not have any of those attributes. It was felt that if P is interpreted as
effort-Pass junction and S as effort-Sum junction, the letters S and P could still be
retained. It was also suggested that P-junction may be called a Pascalian implying
undiminished passage of effort to all bonds appended to it.

Fig. 1.7(a) Fig. 1.7(b)

However, no matching term could be associated with S-junction. A Sanskrit term


‘Samagatik’1 may perhaps be justifiably used to support S-junction along with
‘Paribalak’2 for P-junction. However, one of us had a long discussion with Dean
Karnopp (one of the early associates of the originator of Bond graphs). It was agreed
that as far as possible a common set of conventions should be used world over. From
here onwards following convention would be mostly followed in this book, barring a
few instances where we have adapted the materials from our earlier works;

(a) 1 - Would be mostly used to represent common flow junction


(b) 0 - Would be mostly used to represent common effort junction

1.7 Reference power directions on the bonds

When we analyze a problem of mechanics, say,


the problem of a single mass and spring system
as shown in Fig. 1.8, we fix a coordinate system.
We may take displacement, x, and all its time
derivatives positive towards right. The force
acting on the mass may also be taken positive
towards right. The system, however, in the
course of motion may attain such a state that Fig. 1.8.
when it is displaced towards the right, the force
on the mass happens to be towards the left. This phenomenon may be interpreted
from the solution of system equations when a positive value of displacement and a
negative value of force are obtained.

1
equal or same motion
2
distributor of effort
13 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

So, initial fixing of positivity is arbitrary, however, further analysis is relative to this
fixation. In practice, bond graphs are drawn for general systems. Thus left and right,
up and down, clockwise or counter-clockwise are of no relevance. One has to then
create a general view point from which any particular system interpretation should
be easily discernible. This is done by assigning the bonds with "Reference power
directions". Which may be as arbitrary as fixing coordinate systems in classical
analysis?

Consider two examples, which are parts of some system models, as shown in
Fig.1.9(a) and Fig. 1.9(b). Where J stands for a junction and E for an element. The half
arrows represent the assigned reference power directions.

Fig. 1.9(a) Fig. 1.9(b)

Through such an assignment the modeler indicates that he has deployed such
orientation of his effort and flow measuring devices or scales that when both these
factors of power are measured positive (or both are measured negative) the actual
power flows in the direction of the half arrow. However, if one of these factors
measures positive and other negative then the actual flow of power is in the opposite
direction.

Thus unlike assigning reference coordinate directions in classical system analysis


where positivity of reference coordinates and efforts are nearly always presumed co-
oriented, in bond graph a conceptual set of two independent scales are used, one of
them measures the flow and other the effort.

Let a simple electrical system be


considered first. The system is made
up of a cell and a resistor as shown in
Fig.1.10 (a). Its bond graph is shown in
Fig. 1.10(b). Let us say that we have
fixed the power directions as shown in
the Fig. 1.10(b).

Fig.1.10(a) Fig.1.10(b)
INTRODUCTION TO BOND GRAPHS 14

This means that power variables are chosen such that, when both acquire positive
values, the power goes from the system to the resistor and to the thermal domain
thereafter. It is known that a flow of current in a resistor creates a drop of voltage
across the terminals with current leaving the terminal at a lower potential. The
product of voltage drop and current gives the power dissipated. So, on the resistor
we have to put an ammeter and a ‘voltage-drop’ meter which could be an ordinary
voltmeter whose negative or ground terminal is connected to the current leaving
terminal of the resistor as shown in Fig. 1.11.

Fig. 1.11 Fig. 1.12

In an experiment with the ammeter and the voltmeter connected as shown in


Fig. 1.11, the characteristics of the resistor may appear as plotted in Fig. 1.12.

What about the cell? From system's view point, if the potential difference across the
terminals of the cell is positive and a positive current flow through it, then the power
comes from cell to the system as shown in Fig. 1.13 below. Same remains true when
both, potential difference and current measure negative.

Fig. 1.13(a) Fig. 1.13(b)

When the entire circuit of Fig. 1.10 (a) is reconsidered with the scheme for measuring
power variables of the two elements as shown in Fig. 1.14(a), the reference power
directions may be represented as shown in Fig. 1.14(b).
15 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

Fig. 1.14(a) Fig. 1.14(b)


So, one may say that behind each assignment of reference power direction to a bond
in a model there is (at least in concept) a scheme of measuring the power variables in
the modeled physical system. Thus unlike classical approach, where a single
reference direction renders positivity of both flows and efforts, in bond graph two
separate scales are used.

Now the question may be asked regarding the reference power direction to be
assigned to the internal bonds. We consider the example of Fig.1.4 (a) and it's bond
graph model as shown in Fig. 1.4(b). We redraw the system with a two alternative
scheme for measuring power variables as shown in Fig. 1.15(a) and Fig. 1.15(b). The
change in the measurement scheme is shown within the dotted circles. The ammeters
and voltmeters indicated by identical suffixes constitute a pair to measure a
conjugate of flows and efforts.

Fig. 1.15(a) Fig. 1.15(b)

Let us look at the happenings (not real but conceptual) from the view points of two
observers A and B positioned at the two versions of the circuits given in Fig. 1.15(a)
and Fig. 1.15(b).
INTRODUCTION TO BOND GRAPHS 16
If the observer A sees that ammeter A3 and voltmeter V3 both read positive, obviously
the power is being drawn from the part of the system on the left of the line ab and
flows towards right of that line. Same is true if both A3 and V3 read negative. Bond 3
of Fig. 1.16(a) records this observation. Where as, if the observer B sees his A3 and V3
both reading positive then the power is drawn from the right of the line ab and flows
to the left of that line. Bond 3 of Fig. 1.16(b) records this alternative. The same effect
may, however, be achieved by reversing the connection of ammeter A3 keeping the
polarity of V3 unaltered. In any bond graph model such an underlying scheme of
instrumentation exists though the scheme may depend on the energy domains
involved in the model.
The reader may find the situation extremely messy as he has to always think in terms
of two scales or a pair of devices. In practice, however, all these arguments are not
needed. In a step by step creation of bond graph model the power directions get
nearly automatically assigned by the very logic of modeling. We recommend the
reader now to read Chapter - 4 of this book where various step by step procedures
are discussed with several examples. The message of reference power directions will
also become clear once the reader goes through that chapter.

The way a change in reference orientation of coordinate axes does not alter the
dynamical behaviour of a system; the reference power direction also does not change
its dynamical behavior.

Fig. 1.16(a) Fig. 1.16(b)

1.8 Bond graph standard elements

In bond graphs, one needs to recognize only four types of basic symbols, i.e., three
single or multiport passive elements, two basic active elements, two basic two port
elements and two basic junctions. The basic variables are effort ( e ), flow ( f ), time
integral of effort ( p ) and the time integral of flow ( Q ).

1.8.1 Basic elements of bond graphs


17 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

Some of the bond graph elements were imported informally earlier


to provide the first impression of a bond graph model. In the
following sections we introduce them in a more organized manner.

The concept of energy ports: In discrete system modeling the basic


elements are incorporated in the system through what may be
called terminals of the element. It may be assumed that no other
parts of the element come in contact with other elements of the
system.

For example the system shown in Fig. 1.17 has three elements
Fig. 1. 17
namely an inertia, a spring and a viscous damper and two
sources, a source of effort F(t) and that of flow V(t).

Considering the physical construction of the system each of these elements has
terminals as shown in Fig. 1.18.

Fig. 1.18

a and b are the terminals of the inertial element. Likewise b and c are that of the
spring and the damper.

In bond graph modeling the inter connection takes place through abstract entities
called energy port. Each energy port is represented by a bond having associated with
it two conjugate factors of power. Say, for example, power can enter or leave an
inertial element through the net force acting on it and the velocity at which it moves.
Thus a simple mass point in one dimensional linear motion is associated with one
pair of power variables and consequently with one energy port.

The compliant element of Fig. 1.18, though has two terminals, may be associated with
only one pair of energy variable and hence one energy port. In this case the energy
variables are the relative velocity between the physical terminals and the force
applied on them (which are equal though opposite). Same is true for the linear
damper of Fig. 1.18.
INTRODUCTION TO BOND GRAPHS 18
In a bond graph model these elements would be represented as shown in Fig. 1.19.
Here J stands for any junction 1 or 0 or alternatively, S or P.

Fig. 1.19

The sources as members of the set of basic elements of bond graphs are also
associated with energy ports. The source of effort has associated with it a prescribed
function of time (or even the states of system) as effort; the flow variable associated is
determined by the system which remains a paradynamic variable. The source itself
does nothing with the flow variable. The bond graph representation of an effort
source would be as shown in Fig. 1.20(a).

Fig. 1.20 (a) Fig. 1.20 (b)

On the other hand, the source of flow has a function of time (or even states)
associated with it as flow variable of the energy port. The effort variable is
determined by the system with which the source does nothing of our interest. A
source of flow is shown in Fig. 1.20 (b).

The junction elements are not something physical. As already discussed, they
represent the connectivity. Any number of elements may be attached to them
including other junctions, thus in bond graph models the junction elements appear as
multienergy ports or simply multiports.

The bond graph model of the system shown in Fig. 1.17 will appear as shown in
Fig. 1.21. It may be observed that the elements, which in the physical model had two
terminals but single pair of energy variables associated with them, have, in the bond
graph model, single bonds attached to them.

How this bond graph model is arrived at is discussed in the chapter entitled “Art of
creating system bond graphs.”
19 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

The elements I, C and R may have their


multienergy port versions as well. Say, a
mass point moves in space. The net
energy stored in it or retrieved from it
depends on three components of force
vector and those of the velocity vector. If
this mass point is treated as a single
entity and not three separate inertial
elements then in bond graph
representation it will be associated with
three energy ports or three bonds as
shown in Fig. 1.22.

Fig. 1.21

In some cases there may even be cross coupling amongst the energy ports. Ji are
junction elements Ji є (1, 0). Similarly, there are multiport versions of compliant and
dissipative elements also as shown in Fig. 1.23(a) and Fig. 1.23(b).

Fig. 1.22 Fig. 1.23(a) Fig. 1.23(b)

Such multiport elements are called fields. In this book we have included discussion
on fields in Chapter – 3.

There are two special elements we have not yet come across. These elements have
two energy ports. The elements are such that they do not change the energy content
of the system at any point of time (somewhat like junction elements). These elements
are transformers and gyrators as shown in Fig. 1.24(a) and Fig. 1.24(b).

Fig. 1.24(a) Fig. 1.24(b)


INTRODUCTION TO BOND GRAPHS 20
The reference power directions are oriented as if power flows through these
elements. In the following section we would discuss the general characteristics of
each of these elements.

1.8.2 Constitutive laws of single port bond graph elements

In this section we will discuss basic constitutive laws of the bond graph elements
with single energy port and the conservative two ports (TF and GY) discussed in the
previous section. We call them basic as the actual forms of them demanded by the
bond graph model for generation of system equation is determined by information
orientation of these elements or the so called notation of causality.

The source of effort (SE): The source of effort is external to a system in the sense that
the system does not influence its output. The source of effort determines the effort in
the bond associated with it. Flow to a source is an extra system entity and no
cognizance is taken of it. The effort prescribed by such a source may be a function of
time and even the states of the system or both. The source functions have restrictions
which are very basic to solvability of system differential equations. Broadly stating
the functions must be Riemann integrable, i.e., they must be piece wise continuous
with discontinuities at countable number of time points. Nearly all physical
situations remain confined to such a class of functions.

The source of flow (SF): This source determines the flow variable in the bond
associated with it. For this source the effort variable is an extra system entity and
does not influence the output of this source. The flow prescribed by such a source
may be any function of time or states of the system or both. The restrictions being
same as stated above for the source of effort.

For book-keeping of the system bond graph model it is advantageous to write the
symbolic form of the function near the mnemonic of the source element used in the
model as shown in Fig. 1.25.

Fig. 1. 25

The pm and Qn in the arguments indicate that the corresponding functions are
functions of system state as well.
21 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

The inertial element (I): Creation of this element is motivated by the idea of a mass
point in mechanics. This element relates effort or its integration (the impulse) to the
rate of change of momentum or the velocity. The relations for a simple linear inertial
element may be written in the following forms.

d (m(t ) f )
e t   , (1.8)
dt
1 t
or f t   e   d . (1.9)
m t  
In these expressions the generalized inertia
m(t) is taken as function of time. In Newtonian mechanics of a particle the mass does
not change with time. However, in rotary motion or in electrical circuits the
generalized inertia varies with time. The expressions above include such
eventualities.

In bond graph theory, momentum of the inertial element plays a central role. In
terms of momentum the above expression may be written as

d  p
e t   (1.10)
dt

p t   e    d
t
 
(1.11)

Which of these expressions is to be used at various stages of generation of system


equations is to be entirely decided by the causal orientation of the inertial element.
Detailed discussion on the concept of causality is given in the next chapter.

The compliant element (C): The compliant element’s rolemodel is a Hookian spring
or an electrical capacitor, which stores energy depending on their configurations. The
basic constitutive relation of a single port compliant element in bond graph theory
may be written in following forms.

 t 
e(t )  F  t ,  f    d  ,
 
 

(1.12)

dG t , e t  
f t  , (1.13)
dt
INTRODUCTION TO BOND GRAPHS 22

where F and G are single valued functions of their arguments. G is of course


differentiable with respect to its arguments. In bond graph theory the integration of
flow is called generalized displacement or charge of the C element,
t
Q(t )   f    d .


The above relations may thus be written as

e t   F  t , Q  t   , (1.14)

dQ t  dG  t , e t  
or  , (1.15)
dt dt

or Q  t   G  t , e t   , (1.16)

where G is a general single valued function of its arguments. The linear Hookian
spring thus has a constitutive relation as follows.
t
e t   K  t   f    d  K  t  Q  t  ,


d  e t  / K  t  
or f t   . (1.17)
dt

Likewise for electrical, heat or fluid capacitor the relation would be

1 t
e t    f    d  Q (t )/C (t ) ,
C t  

d  C  t  e t  
or f t   . (1.18)
dt

The variations in compliant parameters K and C in these linear models indicate


fluctuations of the stiffness or capacitance with time. If these parameters are
functions of states of the system as well then the model would become nonlinear.
23 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

One significant idea of classical system dynamics is that of potential force or effort
and potential function. In general a compliant element would be called potential if
the effort determined by it is derivable from displacement gradient of a scalar
potential function. In classical mechanics the attribute potential is associated with
such efforts. There we also associate it with the element which determines potential
force. This is entirely a matter of usage. A single port compliant element can never
become non-potential, as all scalar continuous functions of single displacement may
be derivative of some other function of the same displacement. Non-potential
elements or effort fields would be possible in higher dimensional or multiport
versions of compliant element. A detailed discussion on this aspect is given in
Chapter – 6. Which version of the constitutive relation has to be used in the
generation of system equations is once again determined by its causal orientation.

The resistive element (R): Unlike I and C elements which relate essentially the
integration of effort or flow variables to flow or effort on a bond, the resistive
element, involves no integration and directly relates the effort to flow or vice-versa.
Its role model is Ohmic resistance or a viscous damper.

The generalized constitutive laws for a bond graph resistive element are as follows.

e   f  or f    e .

In their linear forms these may be e = Rf or 1


f  e,
R
where φ and ψ are single valued functions of flow or effort
respectively. The form which should be used in the consequent mathematical model
depends on the causal orientation of the R element.

1.8.3 Constitutive laws of basic 2-port elements

There are only two kinds of two port elements, namely Transformer and Gyrator. The
bond graph symbols for these elements are TF and GY, respectively and two bonds
are attached to each of these elements.

The Transformer: The bond graphic transformer can represent an ideal electrical
transformer, a massless lever, etc. The transformer does not create, store or destroy
energy. It conserves power and transmits the factors of power with proper scaling as
defined by the transformer modulus discussed afterwards.

The meaning of a transformer may be better understood if we consider the following


example. A massless ideal lever in Fig. 1.26(a) is considered. It is also assumed that
the lever is rigid, which means a linear relationship can be established between
power variables at both the ends of the lever
INTRODUCTION TO BOND GRAPHS 24
.

Fig. 1.26(a) Fig. 1.26(b)

From the geometry we have V2 = (b/a)V1.

Conservation of power implies

V2F2 = V1F1 or F2 = (a/b)F1.

In bond graphs, such a situation may be represented as shown in Fig. 1.26(b). The "r"
above the transformer denotes the modulus of the transformer. The reader may note
the arrowed arc over TF representing the transformer element. The oriented arc
indicates the way the modulus is defined. In the given picture the flow fj is r times fi.
We once again emphasize that the modulus is always declared relating flow to flow.
The effort to effort relation acts in the reverse direction as follows.

fj = r fi is the modulus declaration of flow to flow relation


ei  r e j 

1 
e j  ei 
r 
are the alternative forms of effort to effort relation.
or

Often bond-graphers do not use an oriented arc to qualify the modulus. Our
experience is that the incorporation of the oriented arc brings clarity in the model
and once the graph is causal oriented, the modeler realizes whether the modulus by
itself would appear in the mathematical equations or its reciprocal would appear.
This gives a forewarning of plausible singularities in the model. The modulus r may
be function of time or system states or both. If in the course of simulation there is a
possibility of the modulus becoming unbounded then its direct appearance would
lead to singularity. Likewise, if the modulus appears in reciprocal form, its acquiring
zero value would lead to singularity.

Often transformers with fluctuating moduli are indicated by MTFs. However, in this
book we use only TF with an understanding that modulus may be in general
function of time or system state or both. Incidentally, it may also be a constant.

Regarding orienting the moduli of transformers, we would like to add that in many
situations the modeler may find it easier to obtain the modulus in one way than the
25 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

other. Complex mechanisms are good examples of it. If the model ultimately requires
reversed relations, the numerical reciprocation may be automatically affected.

The Gyrator: A transformer relates flow to flow and effort to effort, where as a
gyrator establishes relationship between flow to effort and effort to flow, again
keeping the power on the ports same. The simplest gyrator is a mechanical
gyroscope shown in Fig. 1.27(a).

Fig. 1.27(a) Fig. 1.27(b)

A vertical force creates additional motion in horizontal direction and to maintain a


vertical motion, a horizontal force is needed. So the force is transformed into flow
and flow is transformed into force. Let Izz be the moment of inertia about z axis, x, y,
z angular velocities about the respective axes and Tx , Ty and Tz the torques acting
about the corresponding axis. Then

Tx = Izzzy and Ty = Izzzx

Thus Txx = Tyy, which also implies conservation of power.

Such a situation may be represented as shown in Fig. 1.27(b). The "" above the
gyrator denotes the gyrator modulus. This modulus is always defined from flow to
effort and no sense of direction is associated with it.

ej =  fi , ei =  fj.

This also establishes the conservation of


power ei fi = ej fj. The gyroscope of
Fig. 1.27(a) can thus be represented by
the gyrator shown in Fig. 1.27(c).
Fig. 1.27(c).

Another good example of a gyrator is the electrical DC motor. A permanent field or


externally energised field DC motor and its bond graph model is shown in
Fig. 1.28(a) and Fig. 1.28(b).
INTRODUCTION TO BOND GRAPHS 26

Fig.1.28(a) Fig.1.28(b)

Here i is the current in armature circuit, e is the back emf produced by the machine, 
is the shaft torque and  is the angular speed of the shaft. The modulus of the
gyrator  is the motor characteristic determined by the magnetic field strength of the
poles. In case of externally energized motor this field depends on field current if. The
constitutive relations for the motor are as follows.

 = i and e =   ,

1 1
or i  and   e .
 

Whenever flow belonging to any particular domain of energy is related to effort


variable of the same or any other domain of energy, the bond graph element gyrator
is used to represent such an action. The modulus of gyrator is not oriented. No
oriented arc is needed to indicate the action of the gyrator. This fact would be
immediately apparent if we consider the gyrator of Fig. 1.27(b) which can be written
in two alternative symmetric forms.

ej = fi and ei =  fj ,

1 1
or alternatively fi  e j and fj  ei .
 

1.8.4 Constitutive laws of the 3-port junction elements

In international bond-graphing there are synonyms for each junction. The junction 1
is also called S or JF whereas the junction 0 is also called P or JE. However, we will
keep using 1 and 0.

These junctions are not physical elements. They do not create, destroy or store
energy. In other words, the algebraic sum of powers at a junction is zero. The
name 3-port used for junctions is a misnomer as will be evident from subsequent
discussions.
27 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

Figure 1.29(a) and 1.29(b) shows two multiport 1-junctions with different power
directions. Assuming inward power direction as positive, the constitutive law for
power conservation, for the 1-junction of 1.29(a), may be written as follows.

Fig.1.29(a) Fig.1.29(b)

e1 f1 + e2 f2 + e3 f3 + e4 f4 = 0.

1-junction being a flow equalizing junction,

f1 = f2 = f3 = f4 .
Hence e1 + e2 + e3 + e4 = 0.
Now consider the bond graph shown in Fig. 1.29(b). In this case the constitutive
relations become
e 1 f1 – e 2 f2 + e 3 f3 – e 4 f4 = 0 and f1 = f2 = f3 = f4 .

Thus e1 – e2 + e3 – e4 = 0.

Therefore, 1-junction is governed by the following rule:

The flows on the bonds attached to 1-junction are equal and the algebraic sum of the efforts is
zero. The signs in the algebraic sum are determined by the half-arrow directions of bonds
appended to the junction.

0-junctions have equality of efforts while the flows sum up to zero. If power
orientations are taken positive towards the junction, the constitutive law for the 0-
junction shown in Fig. 1.30(a) becomes
INTRODUCTION TO BOND GRAPHS 28

Fig.1.30(a) Fig.1.30(b)

e 1 f1 + e 2 f2 + e 3 f3 + e 4 f4 = 0

0-junction being effort equalizing junction

e1 = e2 = e3 = e4 .
Hence f1 + f2 + f3 + f4 = 0.

For the 0-junction of Fig.1.30 (b), the flow sum becomes

f1 – f2 + f3 – f4 = 0

Therefore, a 0-junction is governed by the following rule:

The efforts on the bonds attached to a 0-junction are equal and the algebraic sum of the flows
is zero. The signs in the algebraic sum are determined by the half-arrow directions of the
bonds appended to the junction.

1.8.5 Mechanical 1 and 0 junctions

The mechanical 1-junction is straight forward. Elements which have the same
velocity are attached to 1-junction. The mechanical 0-junction needs some
consideration. To understand a mechanical 0-junction, let a massless spring be
considered as shown in Fig. 1.31. Let us say, there are two persons trying to stretch
the spring. The effort applied by the person (1), is, say e1(upwards) and the effort
applied by the person (2) is say e2 (downwards). If the scales chosen are disposed as
shown in Fig. 1.31, then both will read their efforts as positive.
29 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

Fig. 1.31 Fig. 1.32

As the spring is massless and as infinite accelerations are not permitted, we have e1 =
e2. If end (1) has a flow scale aligned to its effort scale and if the same is true for end
(2) then for positive values of the variables (e1, f1, e2 and f2) the persons at both the
ends will give power to the spring. The power should enter the spring. The effort felt
by the spring will be same as e1 or e2 and the flow felt by the spring will be the relative
velocity of its ends. Thus a bond graph as shown in Fig. 1.32 may be drawn. The
effort of the spring is not a force in the sense of the term. In fact it is an ordered set e3
= < e1, e2 > or according to Fig. 1.32 e3 = < e1, e1 > because e2 = e1.

Thus the effort law becomes e1 = e2 = e3. The bond graph in Fig. 1.32 implies f1 + f2 – f3 =
0, or f3 = f1 + f2 which is compatible with what has been just said viz. the flow observed
by the spring is the relative velocity of its ends.

Similarly a damper is a 0-junction as shown in Fig. 1.33.

Fig. 1.33 Fig. 1.34


INTRODUCTION TO BOND GRAPHS 30
We leave it to the readers to interpret the other power dispositions as shown in
Fig. 1.34. A simple mechanical system is shown in Fig. 1.35. For this system, two
alternative bond graphs are given in Fig. 1.36(a) and Fig. 1.36(b). The reader should
satisfy himself about their validity.

Fig. 1.35 Fig. 1.36(a) Fig. 1.36(b)

1.9 Power direction and physical system coordinates

For positive effort and flow or negative effort and flow, the power direction tells us
in which way the power flows. It also tells us what kind of bi-scales are being used
by different elements. The observer or analyst may see the system through junctions,
for one cannot get inside an element.

In what follows, we discuss how a particular coordinate interpretation is obtained


from a power directed bond graph. Without attempting to create any general theory,
we discuss this with the help of an example of a mechanical system. The system is
shown in Fig. 1.37(a). The sources are with double arrows as their directions for
positive values are not yet determined. It is assumed that the positive displacement
of mass and positive force on the mass are directed upwards in the direction of
positive X as shown.

The bond graph for this system is power directed in two different ways as shown in
Fig. 1.37(b) and Fig. 1.37(c). We will use the following notation in analyzing the
b
meaning of the effort variable. Efforts are written as e j ,elem where b is the bond
number, j is the junction and ‘elem’ is the element.
31 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

Fig. 1.37(a) Fig. 1.37(b) Fig. 1.37(c)

b
This means that e j ,elem is the effort applied by the system through bond graph
b
junction j on the element ‘elem’ at the bond b. Similarly e elem , j means the effort
applied by the element on the system through the junction j at the bond b. In this
particular example, the effort changes its direction across a bond. Now we proceed to
analyse the system coordinate (See Table 1.2) as per the bond graph shown in
Fig. 1.37(b).

Table 1.2

Analysis of physical coordinates for bond graph of Fig.1.37(b)

Positive values of variables Direction Remarks


f2  Upwards Assumption
f1, f3  Attached to the same 1-junction
e1SE ,1 F(t)  Arrow head at 1-junction

e11, SE  Downwards Reversal of effort across a bond

e12 , I  Arrow head at I

e13, 0  Arrow head at 0-junction


5
e SF  Equilibrium of spring
.0
f 5, ,V(t)  Downwards Arrow head at 0-junction
INTRODUCTION TO BOND GRAPHS 32
Hence to confirm to the bond graph of Fig. 1.37(b), the system of Fig. 1.37(a) should
have the orientation of external and dynamic variables as shown in Fig. 1.38.
Analysis of the bond graph in Fig. 1.37(c) is given in Table 1.3.

Fig. 1.38 Fig. 1.39

Table 1.3

Analysis of physical coordinates for bond graph of Fig.1.37(c)

Positive values of variables Direction Remarks


f2  Upwards Assumption
f1, f3  Attached to the same 1-junction
e1SE ,1  Arrow head at 1-junction

e 03,1  Arrow head at 1-junction

e 05, SF  Equilibrium of spring


5
e SF  Reversal of effort across a bond
.0
f 5, ,V(t)  Arrow head at 0-junction

The physical system now has the orientation of coordinates and externals as shown
in Fig. 1.39. We leave it to the reader to figure out the meaning of positive flow in
bond 4 in both cases. Often such an analysis may not be necessary for the entire bond
graph as many bond graphs are drawn keeping the physical system coordinates in
mind.
33 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

The procedure outlined here is not general enough to be applicable to systems in all
energy domains. However, one has to evolve some such argument for interpretation
of physical system coordinates and orientation of externals. One may also discover
many short cuts for such an analysis.

You might also like