Chapter 3:

Generation of System Equations

3.1 Selection of system states

In this chapter we discuss the method of generation of system equations. From an

augmented (power directed and causalled) bond graph, using a step by step
procedure, system equations may be generated. The difference between equations
derived from bond graphs and otherwise is that there will be a set of N first order
differential equations, where N is the number of states. The total number of states is
the number of lumped elements (I and C) with integral causality present in a system.

A system or a part of it, changes from one configuration to another because there
have been, or there are causes which act on it, and the system or a part of it absorbs a
part of these causes. The amounts of absorbed causes may be good measures of the
states of the system and the way these absorbed causes are changing with time
contains the description of its dynamics. The system variables in which the equations
are derived are thus the absorbed causes in storage elements with integral causalities.
The spring or capacitor element (with integral causality) receives flow and thus
displacement or total charge is the variable associated with this element as

t
Q  
f dt .

Q is the general symbol which is used for displacement as well as charge. An inertia
or inductance with integral causality receives effort as cause. Thus the system
variable associated with this element is a generalized momentum as given by

t
p 
e dt .

In general it may be said,

t
System Variables for bond graph based analysis = 
(cause) dt, where cause is the
information going to storage elements with integral causality.

3.2 Generation of system equations

The system equations may be generated by answering the following two questions:
 GENERATION OF SYSTEM EQUATIONS 60
1. What do the elements (all) give to the system (expressed in terms of system
variables and sources)?

2. What does the system give to storage elements with integral causality?

These two questions can be answered by looking into the causal structure of the
bond graph. Power direction, however, decides the signs of various state variables
and internal variables entering the equations.

Let us consider a system shown in Fig. 3.1(a). The bond graph for the system is given
in Fig. 3.1(b).

Fig. 3.1 (a) Fig. 3.1 (b)

The storage elements are I2 and C6 and both have integral causality. Thus the system
variables are p2 and Q6.

p2 = ∫ e2 dt and Q6 = ∫ e2 dt

(1) What do the elements give to the system?

p2
(a) Element I2 gives the flow f 2  .
m2
(b) Element C6 gives the effort e6 = K6Q6.

(c) Element R7 gives the effort e7 = R7 f7.

Now f7 is not a system variable. Let us see from where the information f7 comes. The
information f7 comes from the 1-junction, where it is brought by bond 4. Bond 4
receives this information from the 0-junction, where the flow information is brought
by bonds 3 and 5. Thus the whole derivation for e7 may be as follows:
61 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

e7 = R7 f7 = R7 f4 = R7 (– f3 + f5),

 p 
or e7  R7 ( f 2  V (t ))  R7   2  V (t ) 
 m2 
No further analysis is needed as the equation has come in terms of system variables
and external sources.

(2) What does the system give to the storage elements with integral causality?

(a) To I 2, the system gives the effort

e2  p 2 ,
or p 2  e3  e1 ,
or p 2  e3  F  t   e4  F  t   e6  e7  F  t  ,
or p 2  K 6Q6  R7    p2 / m2   V  t    F  t  .

The above equation has been obtained by substituting e6 from 1(b) and e7 from 1(c).
This equation need not be reduced further as it is completely expressed in terms of
system variables and external sources.

(b) To C6, the system gives the flow

f 6  f 4   f 3  f 5   f 2  V (t )  Q 6 ,
or   ( p / m )  V (t ).
Q6 2 2

The above equation may be left in this form as it is already expressed in terms of
system variables and external sources.

The two system equations are

p 2  ( R7 / m2 ) p 2  K 6 Q6  R7V (t )  F (t ).
  (1 / m ) p  V (t ).
Q6 2 2

These equations may be written in matrix from as shown below.

 GENERATION OF SYSTEM EQUATIONS 62

p2
    R7 / m2  K 6   p2  1 R7   F  t  
         

Q6 
   1 / m2  0  Q6  0 1  V  t  
. (3.1)

d
This is of the form  X    A X    B U  ,
dt
Where {X} is the state vector, {U} is the input vector, [A] is the system matrix and [B]
is the input matrix.

3.3 A bond graph with a transformer element

A transformer is an element which gives two different information on two bonds

depending on the causal strokes. Consider a system and its bond graph
representation as shown in Fig. 3.2(a) and (b).

Fig. 3.2 (a) Fig. 3.2 (b)

The storage elements are I1, C2, C6 and I7, and all these have integral causality. Thus
the state variables are p1, Q2, Q6 and p7.

(1) What do the elements give to the system?

p1
(a) I1 gives the flow f1  .
m1
(b) C2 gives the effort e 2  K 1 Q 2 .

(c) C6 gives the effort e6  K 2Q6 .

63 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

p7
(d) I7 gives the flow f7 
m2
 p 
(e) R3 gives the effort e3  Rf3  Rf1   1 .
 m2 
b b b p 
(f) On bond 5, TF gives the flow f5    f 4    f1     1 .
a a  a   m1 
b b b
g) On bond 4, TF gives the effort e 4   e5   e6    K 2 Q6 .
a a a

(2) What does the system give to storage elements with integral causality?

(a) To I1, the system gives the effort e1  e2  e3  e4  e8  p

 1.
 R b
or p 1   K1Q2    p1    K 2Q6  F (t ).
 m1  a
(b) To C2, the system gives the flow
p
  f
f2  Q2 1 or Q 2  1 .
m1
(c) To C6, the system gives the flow
b p p
  f  f
f6  Q6 5 7 or Q 6    1  7 .
 a  m1 m2
(d) To I7, the system gives the effort
e7  p 7  e6 or  7  K 2Q6 .
p

These equations may be written in matrix form as follows.

p1   b  p  1 
     R / m1  K1
a
K2 0  1   
Q2    Q2  0 F (t ).
     1 / m1 0 0 0
  0 
Q6    b / am1 0 0  1 / m2  Q 6 
p   
   p7  
0 

 7  0 0 K2 0  
(3.2)

3.4 Electrical systems

 GENERATION OF SYSTEM EQUATIONS 64
In electrical systems, the causes to integrally causalled I and C elements are voltage
(e) and current (i) and thus the system variables are generalised momentum
t
p  Li    e dt. and generalized displacement or the charge
t
Q  
i dt .

An electrical circuit and its bond graph are shown in Fig. 3.3.

Fig. 3.3

The system variables are p2, Q4 and p5.

(1) What do the elements give to the system?

p2
(a) I2 gives the flow f2  .
L2
1
(b) C4 gives the effort e4  Q4 .
C
p5
(c) I5 gives the flow f5  .
L1
(d) R3 gives the effort e3  R2 f 3  R2 f 2  R2 ( p2 / L2 ).

(e) R6 gives the effort e6  R1 f 6  R1 f 7  R1 ( f 8  f 9 )  R1 ( f 5  f 2 ) ,

p p 
or e6  R1  5  2  .
 L1 L2 

(2) What does the system give to the storage elements with integral causality?

(a) To I2, the system gives the effort

e2  p 2  e9  e3  e7  R2 ( p2 / L2 ),
65 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

p 
or p 2  ( e6  e1 )  R2  2 ,
 L2 
 p   p 
  E (t )  R2  2 .
p
or p 2   R1  5  2  L 
  L1 L2    2
(b) To C4, system gives flow
 
  f or Q
1
f4  Q 4 5 4 p5 .
L1
(c) To I5, the system gives effort
1 1
e5  p 5  e8  e4  e7  Q4  (e6  e1 )  Q4 ,
C C
p p  1
or p 5   R1  5  2   E (t )  Q4 ,
 L1 L2  C

Thus the state space equations may be written as follows.

  R1 R2  
p2  
 L  L

 0  R1 / L1 

p2  1 
     2 2     
Q4    0 0 1 / L1  Q4   0 E (t ).
p  
    R1 / L2  1/ C  R1 / L1   
 p5  
1 

 5
 
 
(3.3)

3.5 Systems with differential causality

Differential causalities occur in systems having such storage elements of which the
outputs (consequences) are determined by output of some other storage elements or
sources. This, however, does not mean that the parameters of these storage elements
do not appear in the equations. In such cases these parameters get associated with
parameters of other storage elements which have achieved integral causality.

In the following we discuss some systems which have storage elements with
differential causality.

Example 1:
 GENERATION OF SYSTEM EQUATIONS 66

Fig. 3.4(a) Fig. 3.4(b)

The bond graph for the system of Fig. 3.4(a) is drawn in Fig. 3.4(b). The element I6
has differential causality as the flow (and thus momentum) of mass m2 is determined
by flow of mass m1.

Example 2:

Fig. 3.5(a) Fig. 3.5(b)

Element C4 in Fig. 3.5(b) has differential causality as the effort (thus charge) on the
capacitor is determined by the source.

Example 3:

Fig. 3.6(a) Fig. 3.6(b)

67 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

The inertance in the above bond graph has got differential causality as the flow on
the junction is determined by the source.

Now we shall analyze a system with differential causality. The system is shown in
Fig. 3.7(a). The bond graph for this system is shown in Fig. 3.7(b).

The system variables for this system are p1, Q4 and Q8 only (why?).

What do the elements give to the system?

p1
f1  , e4  K1Q4 , e8  K 3Q8 ,
m1
d
F2 (t )  K1 f 4
1 d K
f 5  e 5 / K 2   e 7  e 4  / K 2  dt  F2 (t )  1 f 5 ,
K2 K 2 dt K2
1 d
or f5  F2 (t ).
K1  K 2 dt

Fig. 3.7(a) Fig. 3.7(b)

What does the system give to the elements with integral causality?

 1  e3  e8  F1 (t )   F2 (t )  F1 (t )  K 3Q8. .
p
1 d
Q 4  f 5  F2 (t ).
K 2  K1 dt
p
Q 8  1 .
m1
The above equations may be written in matrix form as follows.
 GENERATION OF SYSTEM EQUATIONS 68
p1   0 0  K 3   p1  1  1
         F (t ) 
Q
 4   0 0 0  Q4   0 0  1 
    0     0   F2 (t ) 
Q8  1 / m1 0  Q8  0 
0 0 
 1   F (t ) 

 0  1 .
 K1  K 2   2 
F (t ) 
0 0 
(3.4)

The third term on right hand side does not appear when a system does not have
differential causality. In general the equations of motion for systems with differential
causality may be written as

d
 X    A X    B  U    E  d  U . (3.5)
dt dt

In some cases of course all entries of [E] matrix may be zero. Say, in the system
shown in Fig. 3.7(a), if K1 spring were not there all entries in [E] would be zero. In fact
nonzero [E] matrix is generated in a bond graph with differentially causalled
elements when in a junction the strong bond is a differentially causalled element.

3.6 Activation and activated bonds

Apart from portraying exchange of power (along with information), model of a

dynamic system often demands portrayal of purely information exchange. So far in
this book a fully augmented (power directed and causal oriented) bond represented
exchange of power as well as information of flow and effort. Pure exchange of
information is represented by activation of a bond. A bond may either be flow
activated or effort activated. An effort activated bond imparts only effort information
to the system at the port which is effort causalled (the end with a stroke). Such an
effort activated bond does not take cognizance of flow information. Likewise a flow
activated bond imparts only flow information to the system at the port which is flow
causalled (the unstroked end) and does not take any cognizance of effort
information. In a power bond both the signals are activated. Though this phrase is
never used.

Perhaps a better starting point to discuss the idea of activation would be to consider
those elements which are inherently activated, the way they are defined. A source of
effort, SE, keeps imparting information of effort to the system without taking
cognizance of the flow information of the bond. Thus one may say that source of
69 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

effort, SE, is inherently effort activated. Similarly a source of flow, SF, keeps
imparting flow to the system without taking cognizance of effort information and the
bond may be said to be inherently flow activated.

The summary of the above discussion is that activation of a signal, either effort or
flow, on a bond means that only cognizance of that signal is taken in the dynamics of
the system. Two inherently activated situations are,

(a) the bond attached to a source of effort, SE, is effort activated, and
(b) the bond attached to a source of flow, SF, is flow activated.

To represent activation of bonds attached to source elements no additional notations

are needed.

We now develop notations to represent activation of other bonds. Let us consider a

fully augmented bond in a model as shown in Fig. 3.8(a). We wish to effort activate
this bond so that it imparts only the effort information. One scheme would be to
exploit the inherent effort activation of the SE element as shown in Fig. 3.8(b).

Fig. 3.8(a) Fig. 3.8(b) Fig. 3.8(c)

Here the bond is replaced by a source of effort which is modulated by the effort
information in the bond. The 0e-junction signifies that effort information is tapped at
this junction and the line with full arrow signifies the signal by which the SE element
is modulated.

This notation is too elaborate. A concise notation may be used to represent this idea
as shown in Fig. 3.8(c). A full arrow is added somewhere within the bond signifying
that the bond is an information bond and not a power bond and the letter e written
near it indicates that it is an effort activated bond. In this representation all the
aspects of an augmented bond are retained.

In the same way say, the bond of Fig. 3.8(a) is to be flow activated. One may use an
elaborate scheme to represent it as shown in Fig. 3.9(a). However, a concise
representation would be as shown in Fig.3.9 (b).
 GENERATION OF SYSTEM EQUATIONS 70

Fig.3.9(a) Fig.3.9(b)

The full arrow is only a reminder that the bond is an information bond (or an
activated bond). The letter f written near it indicates that the bond is flow activated.
In this book only the concise notations have been used to represent activation or
activated bonds.

There is another context where activated bonds are sometimes used in the literature.
In nonlinear systems representation one may vary the moduli of two port elements
as functions of state variables, flow or effort and/or time as shown in Fig.3.10 (a) and
Fig.3.10 (b).

Fig.3.10 (a) Fig.3.10 (b)

The full arrows indicate the signals that modulate the two port elements (TF, GY). In
this book we will not use this notation. Modulation of two ports will be represented
simply by writing the moduli as functions of state variables, flows or efforts or time
or all. This will keep our notations concise as shown in Fig. 3.11(a) and Fig.3.11 (b).

Fig.3.11 (a) Fig.3.11 (b)

3.6.1 Power associated with activated bonds

Power associated with those activated bonds which impart the information of
modulations of two ports as shown in Fig. 3.10 is zero. These bonds are indeed,
purely information bonds, something like modulating amplifications of amplifiers.
Such activated bonds have not been used in this book to represent modulating
signals.
71 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

The power associated with a general activated bond needs some discussions.
Consider an effort activated bond as shown in Fig.3.12 (a).

Fig.3.12 (a) Fig.3.12 (b) Fig.3.12(c)

The effort causalled end of this bond imparts effort information to the J1-junction and
acts like a source of effort. It may thus impart any amount of power as demanded by
the system to accommodate this information. As the flow signal is not activated, it is
forbidden to play any role in the dynamics of the system. Thus the power imparted
by this bond at the J2-junction is zero. Such an activated bond does not conserve
power. It may create (or absorb) power. The question then is from where does this
power come? In fact an effort activated bond may be considered to be an effort to
effort (say, voltage) amplifier with unit gain as shown in Fig. 3.12(b). All such
amplifiers have a power supply system (may be called a tank). The output power is
derived from that tank. The input impedance being very high (infinite in ideal case)
the input power is small (zero in ideal case).

Similarly a flow activated bond as shown in Fig. 3.12(c), the effort causalled end is
power less whereas the flow causalled (or unstroked) end impart any power
demanded by the system at that port.

3.6.2 System equations for models with activated bonds

Let us consider the spring-mass system shown in Fig 3.13(a). The velocity pickup
observes the velocity of the block of mass m which is amplified by a power amplifier
and drives a force actuator which imparts force proportional to the velocity to the
block. The net amplifier gain is being μ. This device can be modeled in two ways.
One using a modulated source of effort as shown in Fig.3.13 (b). Alternatively using
activated bonds as shown in Fig.3.13(c).

The velocity to force (flow to effort) amplifier is modeled with the help of a gyrator
element with its ports suitably activated.
 GENERATION OF SYSTEM EQUATIONS 72

Fig.3.13 (a)

Fig.3.13 (b) Fig.3.13(c)

We now derive the equations of motion for the system.

(1) What do the elements give to the system?

p1 R p
f1  , e3  KQ3, e2  p1 and e5   f 4  1  .
m m m

(2) What does the system give to elements with integral causality?

e1  p 1  e 2  e3  e5 ,
R 
or p 1   p1  KQ3  p1.
m m
1
and Q 3  p1 .
m
Above equations may be written in matrix form as follows.
73 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

1 
 p 1   m    R   K   p1 
  1   . (3.6)
Q3   0  Q3 
 m 
Let us consider another system as shown in Fig. 3.14(a). A DC motor is driving a
fluctuating resistive load through a shaft of torsional stiffnesses K t1 and K t2 and
with two disks. The speed, picked up by a tachometer is fed back to control the speed
of the motor by adjusting the armature current (increasing voltage alternatively). The
bond graph for the system is drawn as shown in Fig. 3.14(b).

Fig. 3.14(a)

Fig. 3.14(b)

Inertia or compliant elements with activated bonds may be used as observers for
recording displacement, velocity or effort. Suppose an 1-junction denoting common
flow does not have a C element attached with it. The flow or the displacement at the
1-junction is therefore not a system state. However, to generate the information of
flow, one can attach an additional C element to this junction, which will record the
flow and its integral (displacement). The value of the stiffness should be given as
zero, alternatively, the bond connected to the C element may be flow activated.
Similarly to measure effort at a 0-junction, which is not a state, an I element may be
 GENERATION OF SYSTEM EQUATIONS 74
attached. Since the value of the inertia for this I element cannot be given an infinite
value, the only alternative is to effort activate the connecting bond. All these
additional elements are termed as observers and the additional states thus created are
called observer states. These states do not appear on the right hand side of any state
equation, and hence do not change the dynamics of the system in anyway. They are
added only for the purpose of observation.

3.7 Fields

So far the external elements like C, I and R were connected to a single bond like –C, –I
and –R. The single port C, I and R elements may also be generalized. To start with, let
us consider the multiport generalization of element C. Whenever the efforts in a set
of bonds are determined by displacements in the bonds of the same set as

n
ei   K ij Q j , i  1,...., n
j i

the relation may be represented by a

multiport C element called C - field.
Similar relationships can be established
for I and R elements as well. The
parameter fields are always referred
enclosed within square braces ([K], [I]
and [R]) as matrices. As an example a
simple situation shown in the Fig. 3.15
may be considered.
Fig. 3.15

Two very long springs of stiffnesses Kxx

and Kyy are attached to a mass m. The
springs are oriented in X and Y
directions. In X and Y coordinates there is
no cross relation between the efforts in X
and Y direction and corresponding
Fig. 3.16
displacements. However, if this problem
is to be solved in X’ and Y’ frame, the
effort along any direction is related to displacements along both the axes.

Then the equations may be written as

75 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

 Fx '   cos sin    K xx 0  cos  sin    X '

    
 Fy ' 
  sin 
 cos   0 K yy   sin  cos   Y ' 


 Fx ' 
  K x' x' K y ' y '   X '
    ,
K y' y' 
or
 F
 y' 
 K y'x'  Y ' 
(3.7)

where K x ' x '  K xx cos 2   K yy sin 2 

K x ' y '  K y ' x '  (  K xx  K yy ) cos sin 
K y ' y '  K xx sin 2   K yy cos 2  .

Dynamics of the mass point can be represented by the bond graph shown in Fig. 3.16.

Bond 3 and bond 4 represent the velocities in the X’ and Y’ directions respectively.
There are direct and cross stiffnesses in the matrix of the two-port C - field.

If the system also had dampers as shown in Fig. 3.17, the dynamics of the mass point
m could be represented by the bond graph shown in Fig. 3.18. The damping may now
be depicted by the R field as shown in the figure. Fields could have been avoided by
selection of multiple coordinates for springs and dampers and connecting them with
four transformers for each.

The field elements represent the spring and damping with cross-couplings. Such a
2×2 field may be rotated to bring it to a diagonal form (not always however!), which
correspond to the principal directions. Journal bearings are modeled using 2×2 C and
R fields. Transistors are modeled using 3×3 or 2×2 R fields.
 GENERATION OF SYSTEM EQUATIONS 76
Fig. 3.17 Fig. 3.18

Occurrences of [C] - fields are common in analysis of beam vibration problems,

where the basic beam element is represented by a 4x4 stiffness matrix. This matrix
relates the two sets of bending moments and shear forces at both ends of a beam
element to the corresponding set of angles and displacements.

[C] - fields are a common feature in modeling of thermodynamic systems. For

instance, a collapsible chamber in an engine or a compressor chamber can store
energy through interaction of three modes, viz. the mechanical port associated with
the piston, thermal port for the heat transfer and the chemical work done by mass
transfer and combustion.

We consider another system as shown in Fig. 3.19. Here the coordinate x1 is not
inertial as it is the displacement of mass m1 measured from m2. The effective masses in
these coordinates may be obtained from the expression of kinetic energy ( T ) and
deriving generalised momentum ( p ) associated with x1 and x2 coordinates.

1 1
T  m1 ( x 1  x 2 ) 2  m 2 x 2 2 .
2 2

The generalized momenta may be derived as

follows.

 T 
p x1     m1 ( x1  x 2 ).

 x1 
 T 
p x 2     m1 ( x1  x 2 )  m2 x 2 .

 x 2 
Fig. 3.19
These relations may be written in matrix form as

 p x1   m1 m1   x 1 
   (3.8).
 p x 2   m1 m1  m 2 
  x 2 

The bond graph of the system may be

drawn as shown in Fig. 3.20.

Here the generalized momenta with the

cross terms as given in the above matrix
Fig. 3.20 equation is depicted by an I - field.
77 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

In mechanical systems, I - fields occur

whenever noninertial coordinates are
used (e.g. modeling a rotating body in a
three dimensional space frame).

3.7.1 System equations of motion with field elements

Consider the bond graph shown in Fig. 3.18. The state variables are p1, p2, Q3 and Q4.
The constitutive relations are
p1 p2
f1  , f2  , e3  K x ' x ' Q3  K x ' y ' Q4 ,
m m
e 4  K y ' x ' Q3  K y ' y ' Q4 ,
1 1
e5  ( R x ' x ' p1  R x ' y ' p 2 ) and e6  ( R y ' x ' p1  R y ' y ' p 2 ).
m m

The state equations are

p 1  (e3  e5 ),

or p 1   K x ' x ' Q3  K x ' y ' Q4 

1
m
  
R x ' x ' p1  R x ' y ' p 2 .
 
p 2  (e4  e6 ),

or  2   K y ' x ' Q3  K y ' y ' Q4 

p
1
m

R y ' x ' p1  R y ' y ' p 2  .
 

  p1 ; Q
Q   p2 .
3 4
m m
Or in matrix from

p1   R x'x' / m  Rx' y' / m  K x'x'  K x ' y '   p1 

    
 p2   R y 'x' / m  R y'y' / m  K y 'x'  K y' y'  p2 
    
  .
Q3  
1/ m 0 0 0

Q3 
Q  
 0 1/ m 0 0  
 Q4 
 4  
(3.9)

3.8 Algebraic loops

 GENERATION OF SYSTEM EQUATIONS 78
Let the bond graph for a particular
system be considered as shown in
Fig. 3.21. The strong bonds for three
junctions are connected with R2, R6
and R8 respectively. Equations for the
junction variable are not directly
determinable, since only storage
elements define integral
relationships. The state variables here
are Q3, p7 and Q9. Fig. 3.21

The constitutive relations may be written as

e2 e
e3  K 3Q3 , f 7  p7 / m7 , e9  K 9Q9 , e6  R6 f 6 , f2  and f8  8 .
R2 R8
The state equations are
e 1
Q 3  f 3  f 2  2   SE1  K3Q3  R6 f6 . (3.10)
R2 R2
 p 
p 7  e7  R6 f 6  R6  f 4  f 5  f 7   R6  f 2  f8  7  . (3.11)
 m7 
e 1
Q 9  f 9  8   e5  e9   1  R6 f6  K9Q9  . (3.12)
R8 R8 R8

Now it may be observed from the state equations that in the first equation the term f6,
which is nothing but the contribution of R6 element, is appearing. Similarly, in the
second equation terms f2 and f8, the contributions of elements R2 and R8 respectively
are appearing. f6 also appears in eqn. (3.12). Hence, state equations cannot be written
independently. Such a situation is termed as algebraic loop in bond graph
terminology. The above mentioned algebraic loop is of third order. To resolve such a
situation the equations need to be expressed in terms of state variables. For doing
this, following three methods can be used.

 Expanded Method
 Optimized Method
 Matrix Method

Expanded Method:
e2  SE1  K 3 Q3  R6 f 6 . (3.13)
f 6  e2 / R2  e8 / R8  p 7 / m7 . (3.14)
e8  R6 f 6  K 9 Q9 . (3.15)
79 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

Substituting the expression of e2 from eqn. (3.13) in eqn. (3.14)

f 6   SE1  K 3Q3  R6 f 6  / R2  e8 / R8  p 7 / m7 ,
or 1  R6 / R2  f 6   SE1  K 3Q3  / R2  e8 / R8  p7 / m7 ,
or ID1 f 6   SE1  K 3Q3  / R2  e8 / R8  p 7 / m7 ,
or
f 6    SE1  K 3Q3  / R2  / ID1  e8 / R8 / ID1  p7 / m7 / ID1 , (3.16)

where ID1  1  R6 / R2 .

Substitution of f6 in the expression of e8 leads to

e8  R6    SE1  K 3Q3  / R2  / ID1   R6e8 / R8 / ID1  R6 p7 / m7 / ID1  K 9Q9

,
or
1  R6 / R8 / ID1  e8  R6   SE1  K3Q3  / R2  / ID1  R6 p7 / m7 / ID1  K9Q9
,
or ID2e8  R6   SE1  K 3Q3  / R2  / ID1  R6 p7 / m7 / ID1  K 9Q9 ,
or
e8  R6    SE1  K 3Q3  / R2  / ID1  / ID2  R6 p7 / m7 / ID1 / ID2  K 9Q9 / ID2
, (3.17)
where ID2  1  R6 / R8 / ID1  .

The state equations can now be expressed independently as follows.

p 7  (((SE1  K 3Q3 ) / R2  p7 / m7 ) / ID1  ((((SE1  K 3Q3 ) / R2  p7 / m7 ) / ID1 ) R6

 K 9Q9 ) / ID2 / R8 / ID1 ) R6 .
(3.18)
Q  ( SE1  K Q  ((( SE1  K Q ) / R  p / m ) / ID  (((( SE1  K Q ) / R
3 3 3 3 3 2 7 7 1 3 3 2

 p7 / m7 ) / ID1 ) R6  K 9Q9 ) / ID2 / R8 / ID1 ) R6 ) / R2 .

(3.19)
Q  ((((SE1  K Q ) / R  p / m ) / ID ) R  K Q ) / ID / R .
9 3 3 2 7 7 1 6 9 9 2 8

(3.20)

Note: All the ID (identity) terms created in this method are dimensionless.
 GENERATION OF SYSTEM EQUATIONS 80

Optimized Method: The state equations can further be simplified by creating some
more IDs as defined below.

ID1   SE1  K 3Q3  / R2  p7 / m7 , ID2  1  R6 R2 ,

ID3  R6 ID1 / ID2  K 9Q9 , ID4  1  R6 / R8 / ID2 .

State equations are obtained as follows.

p 7  R6  ID1 / ID2  / ID4 / R8 / ID2 .

(3.21)
Q 3   SE1  R6  ID1 / ID2  ID3 / ID4 / R8 / ID2   K 3Q3  / R2 .
(3.22)
  ID / ID / R .
Q4 3 4 8

(3.23)

Matrix Method: Equations (3.13) through (3.15) can be written in matrix form as
follows.

 1 R6 0  e2   SE1  K 3 Q3 
    
  1 / R2 1 1 / R8   f 6    p 7 / m7 
 1   
 0  R6  e8     K 9 Q9 

 e2   ID 1 
or  R  f 6    ID  2 , (3.24)
 e   ID
 8   3 

 1 R6 0 
 R  / R 1 1 / R8 ,
where =
 2

 0  R6 1 

ID1  SE1  K 3Q3 ; ID 2   p 7 / m7 and ID 3   K 9 Q9 .

From equations (3.10) through (3.12) state equations can be obtained as follows.
81 MODELING AND SIMULATION OF ENGINEERING SYSTEMS THROUGH BOND GRAPHS

Q3  X 1 / R 2 ,
(3.25)
p 7  X  2 R6 , (3.26)
  X  3 / R ,
Q (3.27)
4 8

where  X    R   ID .
1

3.8.1 Mixed causalled fields

Only the case of [R] fields is discussed here. Other fields are complex, because of the
need of partial inversion of field matrices to arrive at the state equations. Three types
of causal patterns are possible in a [R] field, as shown in Fig. 3.22.

(a) (b) (c)

Fig. 3.22

 The first type of causal pattern shows all the bonds causalled with resistive
causality. For such a case, the equations may be written as

 e1   R11 R12 R13 R14   f 1 

e   R R22 R23 R24   f 2 
 2   21
   .
e3   R31 R32 R33 R34   f 3 
e4   R41 R42 R43

R44   f 4 
(3.28)

 When the field is in conductive causality the equations for output variables
may be written as

1
 f 1   R11 R12 R13 R14   e1 
 f  R R22 R23 R24  e 
 2   21  2
   .
 f 3   R31 R32 R32 R34   e3 
  
 f4   R41 R42 R43 R44  
e4  (3.29)
 GENERATION OF SYSTEM EQUATIONS 82

 When the field is mixed causalled, the process of writing the equations is
different. Let [RO] be a unit matrix and [RI] be a matrix containing the
elements of the [R] field, i.e.,

1 0 0 0  R11 R12 R13 R14 

   
0 1 0 0 R R22 R23 R24 
 R0   and  RI    21
 R31
.
0 0 1 0 R32 R33 R34 
   
0 0 0 1   R41 R42 R43 R44 

Now one may proceed as follows.

Interchange those columns of [RO] and [RI], which correspond to conductive

causality in [RI], after making those columns negative. This forms the new [RO]
matrix, from which the equivalent [R], that relates input vectors to output (cause and
effect) may be written as
 R  equiv   RO 1  RI 
1
 e1  1  R12 0  R14   R11 0 R13 0   f1 
      
f  0  R22 0  R24   R21  1 R23 0  e2 
i.e.,  2     .
 R31 0  f 3 
 e3  0  R32 1  R34 
 
0 R33


 f4 
 0  R42 0  R44   R41 0 R43  1 
e4 

(3.30)

It may be noted that in case of complete resistive causality, [RO] = [I], [RI] = [R] and
hence [R] equiv = [R]. In the other extreme case of complete conductive causality, [RO] =
–[R], [R] = –[I] and hence [R] equiv = [R] –1. These two cases satisfy the equations derived
earlier for first two types of causality patterns.