Chapter 9
Introduction to Multi bond graph
9.1 Introduction
Multi bond graphs have the same status in bond graph theory as vectors have in
dynamics. They are very useful in modeling systems where there exist some sort of
morphological similarities in the various subsystem clusters.
In accordance with Breedveld [55, 56, 57, 58, 59], Bonderson's method of
representation of the multi bond, by two parallel lines with one power orientation
half-arrow as shown in Fig. 9.1, is proposed.
The two parallel lines offer the opportunity to
indicate the dimension (number of scalar bonds
represented by a multi bond) by a number written
between the lines. The multi bond graphs are
compact representation of large systems with Fig. 9.1
identical subsystems. Many physical aspects of the
system may get obscured in a multi bond graph. They turnout to be of great value at
the stage when the initial ideas are being formulated. However, it is advisable that a
multi bond graph should be ultimately changed into a usual scalar bond graph for
detailing and simulation. In this book we have not used multi bond graph for
representing system models.
9.2 The various multi bond graph elements
Multi bond graph has its own grammatics of representing system elements, junctions
and transformers or gyrators different from those for scalar bond graph discussed
earlier. This grammatics is discussed below in detail.
9.2.1 The single port elements
Multi bond graph notation for single port elements (SE, SF, I, C, R) is shown below.
In Fig. 9.2, A represents the bond number and n indicates the dimension. The scalar
bond graph equivalents are shown in Fig. 9.3.
INTRODUCTION TO MULTI BOND GRAPH 260
Fig. 9.2 Fig. 9.3
9.2.2 The junction array
Multi bond graph notation for the junction array (1, 0) is shown below. In Fig. 9.4, A,
B and C are the bond numbers and n is the dimension. It should be noted that the
junction array connects only those bonds of the multi bonds coupled by the junction
array, which have a corresponding index. Hence the dimensions of the multi bonds
connected to a junction array have to be identical. The scalar bond graph equivalent
is shown in Fig. 9.5.
Fig. 9.4 Fig. 9.5
9.2.3 Direct sum of multi bonds
Direct sum of multi bonds is introduced
to represent the composition of multi
bonds analogous to the direct sum of
vector spaces in linear algebra. Direct
sum is represented by a transverse line to
the multi bonds which take part in the
summation. Orientation and causality are Fig. 9.6(a)
maintained but the composition (out of
scalar bonds) and order of the multi
bonds may be changed. A direct sum is
shown in Fig. 9.6(a).
The same transverse line may be used to decompose a multi bond into its component
bonds. There are several ways in which a direct sum may be used in a multi bond
graph for composition and decomposition of multi bonds. Figure Fig. 9.6(b) shows a
261 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS
part of a multi bond graph where direct sum is used both ways, for composition as
well as decomposition. The multi bond GY and field elements used in Fig. 9.6(b) are
explained below.
Fig. 9.6(b)
9.2.4 The transformer and the gyrator
Both the transformer (TF) and the gyrator (GY) are in the form of transformation
matrices. They have two ports and each port may or may not have the same
dimension. In a transformer the distributor is a 1-junction and the summer is 0-
junction. The following 32 transformer may be considered as an illustrative
example. The multi bond representation of the transformer is shown in Fig. 9.7 and
the scalar equivalent with distributors and summers is shown in Fig. 9.8.
Fig. 9.7 Fig. 9.8
INTRODUCTION TO MULTI BOND GRAPH 262
In a gyrator both the distributor and the summer are 1-junctions. The following 32
gyrator may be considered as an illustrative example where Fig. 9.9 shows the multi
bond representation of the gyrator and the scalar equivalent with distributors and
summers is shown in Fig. 9.10.
Fig. 9.9 Fig. 9.10
In both the previous examples, all the bonds of the matrices are present. In such a
situation the transformer or gyrator matrix is said to be dense. If some of the bonds
do not exist, then the matrix is said to be sparse.
The only constraint for a sparse matrix is that none of the summers or the
distributors should be de-linked or else the entire bond graph collapses.
9.2.5 Field elements
The field elements (FI, FC, FR) are multi ports. Each port may or may not have the
same dimension. Both the I - fields shown in Fig. 9.11 are possible. The first I - field
has the same dimension n at each port while the second I - field has different
dimensions l, m and n at the three different ports.
Fig. 9.11
263 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS
9.3 Spatial mass spring damper system
Consider a mass, anchored by three spring damper combinations in space in the X, Y
and Z directions and subjected to a set of forces in these directions as shown in Fig.
9.12.
This is a case for vectorization and the multi bond graph for the system is shown in
Fig. 9.13.
INTRODUCTION TO MULTI BOND GRAPH 264
9.4 A semi-definite system
The semi-definite free system shown in Fig. 9.14 can be represented by the multi
bond graph of Fig. 9.15. The field elements C and R are represented by the matrices
K1 K1 R1 R1
and
K1 K1 R1 R1
1 1
If is considered to be the unit symmetric matrix (US), the multi bond graph
1 1
can be represented in a compact form as shown in Fig. 9.16.
9.5 Enhanced semi-definite system
The power of multi bond graphs would be apparent from this particular example.
The system shown in Fig. 9.17 is an extension of the system in Fig. 9.14. The multi
bond graph in Fig. 9.18 is enhanced in its dimension but no further multi bonds are
added.
Fig. 9.17 Fig. 9.18
Even if the system is enhanced to a dimension n the multi bond graph will be
enhanced only in dimension. In this system the field elements C and R become 33
matrices as shown in Fig. 9.18. However, this is not a problem because a linear
combination of unit symmetric matrices with K1 and K2 and concatenation of the
matrices so formed will give the 33 matrix.
9.6 The beam problem
Consider the beam shown in Fig. 9.19. The bond graph for the beam is shown in
Fig. 9.20. At the first glance it does not seem to be amenable to vectorization.
However, a little thought will show that the multi bond graph of Fig. 9.21 is a valid
265 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS
vectorization of the problem. It uses a field C element matrix which is sparse. This is
not a very efficient method. A more efficient method will be to use the concept of
tensorial C element.
Fig.9.19
Fig. 9.21 Fig. 9.20
9.7 Beam with gyroscopic effect
The same philosophy is followed with the system shown in Fig. 9.22 as in the
previous problem. The entire beam is divided into reticules. Each reticule exhibits
morphological similarity with the other. The multi bond graph for this system is
shown in Fig. 9.23. In this case too, the field C element is sparse and an efficient
solution can be obtained by using the concept of tensorial C.
Fig. 9.22
Fig. 9.23
INTRODUCTION TO MULTI BOND GRAPH 266
A program, written in COMMON LISP, which takes the multi bond graph input and
writes down the scalar equivalent has already been developed at I.I.T, Kharagpur.
9.8 Representation of Moduli of Multi-port Transformer and Gyrator
Elements
In scalar context the orientation of TF indicate the flow relation between the ports of
TF as follows
f 2 f1 and e1 e2 .
The relation maintains the conversation of power through the transformer.
Orientation depends the way modeler wants to create these relations and have
nothing to do with power or casual orientation of the ports. For vector bond graph
orientation of modulus pays even more significant role. It relates a vector TF to an
underlying scalar structure without which the model has no sense. Let us consider
two examples when transformer causalities are co-oriented.
Fig. 9.24
The underlying scalar structure is
267 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS
Fig. 9.25
The second example is
Fig. 9.26
Fig. 9.27
INTRODUCTION TO MULTI BOND GRAPH 268
Let analyze the flow and effort relation for the transformer of Fig. 9.24 through the
scalar structure of Fig. 9.27.
f1a
f 2 a a11 a12 a13
f1b (9.1)
f 2b a 21 a22 a 23 f
1c
e1a a11 a 21
e
and e1b a12 a22 2 a (9.2)
e
e a
1c 13 a 23 2b
Let us prove the power conservation relation
e2T . f 2 e1T f1 (9.3)
Substituting for f 2 from eqn. (9.1) and e1 from eqn. (9.2)
a a12 a13 T a11 a12 a13
e2T . f 2 e1T 11 T
f1 and e1 f1 e2 f1 .
a 21 a22 a 23 a21 a 22 a 23
Thus the right side of eqn. (9.3) is equal to the left hand side of eqn. (9.3).
Thus it will be inadequate talking about the modulus of vector transformer with out
assigning the modulating matrix with an orientation.
GYRATOR: Gyrator’s moduli are not orientated. In scalar context the gyration
relation are symmetric.
e2 f1
e1 f 2
For vector gyrators the underlying scalar structure is symmetric, see Fig. 9.28 and
Fig. 9.29.
269 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS
Fig. 9.28
Fig. 9.29
Analysis of a vector transformer when modulus orientation and causality are in
opposite directions, see Fig. 9.30 and Fig. 9.31.
Fig. 9.30
INTRODUCTION TO MULTI BOND GRAPH 270
Fig. 9.31
In creating the underlying scalar bond graph structure as shown in Fig. 9.31, the
orientation of transformer modulus is given prime importance. However, when one
proceeds to causal the graph, one has to retain the causalities of bond 1 (in the scalar
structure bonds 1a, 1b, and 1c) and bond 2 ( in scalar graph 2a and 2b), liberty has to
be taken in assigning causalities to the ports of transformers. The final causal
structures are shown in Fig. 9.31.
However, this leads to creation of causal loops, i.e. the strong bond of junction are
ports of transformers forming a junction loops and in this particular case ports of one
transformer could not be causalled. Thus when causal orientation is in opposition to
the modulus orientation then the vector transformer maybe said to be in improper
causality.
Like wise a gyrator with element causal ports will lead to scalar situation as shown in
Fig. 9.32 and Fig. 9.33.
Fig. 9.32
271 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS
Fig. 9.33
As per the opinion of the reviewer the issue of causality of vector transformer and
gyrator may be better understand if one keeps the underlying scalar structures in
mind. The scalar structures are entirely decided by the orientation of the moduli
matrices in case of vector transformers.