Hydraulics

Lecture Notes in
Control and
Information Sciences
Edited by A.V. Balakrishnan and M.Thoma
33
Peter Dransfield
Hydraulic Control Systems -

Design and Analysis of Their
Dynamics
Springer-Verlag
Berlin Heidelberg New York 1981
Series Editors
h~ V. Balakrishnan - M. Thoma
A d v i s o r y Board
I D. Davisson A. G. J. MacFarlane. H. Kwakernaak
J. I Massey Ya. 7_ Tsypkin A. J. Viterbi
Author
Peter Dransfield, Ph.D.,
Department of Mechanical Engineering
Monash University, Australia
ISBN 3-540-10890-4 Springer-Vertag Berlin Heidelberg NewYork

ISBN 0-387-10890-4 Springer-Verlag NewYork Heidelberg Berlin
This work is subject to copyright. All rights are reserved, whether the whole
or part of the material is concerned, specifically those of translation, re-
printing, re-use of illustrations, broadcasting, reproduction by photocopying
machine or similar means, and storage in data banks.
Under 54 of the German Copyright Law where copies are made for other
than private use a fee is payable to 'Verwertungsgesellschaft Wort', Munich.
Springer-Verlag Berlin Heidelberg 1981
Printed in Germany
Printing and binding: Beltz Offsetdruck, Hemsbach/Bergstr.
906113020-543210
PREFACE
The Text centres around the ideas that for hydraulic control systems
the quality of their dynamic response is important to their proper operation,
prediction of dynamic response capability is best obtained via the

development of dynamic models,
the inability of many designers to efficiently effect predictive dynamic

analysis has its roots in a lack of confidence in the forming of models,
the use of p o ~ bon~ ~aphs is a natural and superior approach to the

development of reliable dynamic models,
digital simulation provides the most appropriate method for extracting

response information from dynamic models.
Among these, the use of bond graphs (abbreviation for power bond graphs) is paramount
to the Text. Therefore, bond graphs are introduced, explained, demonstrated, and
utilized in substantial detail. The features of bond graphs which make them
desirable in the hydraulic control system design situation include their structural
affinity with the real system and its components, the formality in the assembly of a
bond graph structure, their modular and re-usable nature, and the provision of a
structure from which the set of equations which become the model can be formally
prepared.
It is assumed that the reader has a general appreciation of the nature and
operation of hydraulic control systems and their major components. The Text is
written mainly for those mechanical and control engineers who would like to be able
to include predictive dynamic analysis among their techniques for the design of
hydraulic control systems but who find existing approaches unsatisfying or
inappropriate. The Text should be suitable also for up to twenty hours in
appropriate graduate or senior undergraduate courses, and for specialized extension-
type courses.
For full appreciation of the Text, the reader will need to make a "break through"
by thoroughly understanding and using the symbols and structures of bond graphs.
Such a break through was literally forced on me by a graduate student whom I was
supervising for a research Master degree. I gladly acknowledge my debt to
Bevis Barnard, Senior Lecturer at Caulfield Institute of Technology, for the gentle
persuasion emanating from his own discovery and conviction that bond graphs offered
something unique to the modelling, simulation and design of powered control systems.
I acknowledge also the conceptions of Henry Paynter, the original and sustained
developmental work of Dean Karnopp and Ron Rosenburg, the work of bond graph
entrepreneur Jean Thoma, the assistance of colleague Jacek Stecki, and the
contributing work of a small but steady stream of undergraduate and graduate students
including Rob Winton, M.K. Teo, Roger LaBrooy, Kevin Duke, S. Ramachandran, V.K.L. Mai,
and Merren Cliff.
Many people contribute to the production of a text. I single out two for
special thanks and appreciation; Lorry Ryan and John Millar, both of Monash
University, for their excellent work in typing and drafting respectively.
And there is an individual without whose encouragement this Text would not have
been completed my wife Nevillie whose support was sustained, inspirational, and
essential.
To those willing to learn, and to those willing to try a new way, welcome aboard.
P.D. March 1981

CONTENTS
PREFACE
CHAPTER 1 INTRODUCTION 1
1.1 Introduction 1
1.2 A Design Philosophy 2
1.3 Dynamic Modelling 7
1.4 A Desirable Background ii
1.5 Arrangement of Text ii
1.6 Conclusion 12
C~PTER2 NOTATION AND UNITS 13

2.1 Introduction 13
2.2 Notation 13
2.2.1 Symbols 13
2.2.2 Examples of Symbol Use 15
2.2.3 Operations 15
2.2.4 Equations 16
2.2.5 Comments 16
2.3 Units 16
2.3 .i Comments 18
2.4 References 18
CHAPTER 3 CONVENTIONAL MODELLING PROCEDURES 19

3 .I Introduction 19
3.2 Conventional Model Forms 19
3.2.1 Informal Equation Set 21
3.2.2 Transfer Function 21
3.2.3 Vector-Matrix Models 22
3.2.4 Block Diagrams and Signal Flow Graphs 22
3.3 Deriving the Model 22
3.4 Selection of Relationships 24
3.5 Selection of Parameter Values 24
3.6 Another Example 25
3.7 Conclusion 28
3.8 References 28
CHAPTER 4 POWER FLOW MODELLING 29

4.1 Introduction 29
4.2 Power Ports 32
4.3 Describing Power Flow 36
4.4 Getting Equations 40
4.5 An Example 44
4.6 Summary 46
4.7 References 47
CHAPTER 5 POWER BOND GRAPHS 48

5.1 In troduction 48
5.2 Bond Graph Terms and Symbols 49
5.2.1 Effort and Flow Variables 49
5.2.2 Sources 50
5.2.3 Power Bonds 50
5.2.4 Power Transformers 51
5.2.5 Dynamic Effects 54
5.2.6 Resistive Power Dissipation 54
5.2.7 Capacitive Power Storage 57
5.2.8 Inertive Power Storage 58
5.2.9 Summing Junctions 60
5.2.10 Summary of Basic Terms and Symbols 62
CHAPTER 5 (contd.}
5.3 Forming Power Bond Graph Structures 62
5.3.1 Inertia Load with Friction 62
5.3.2 Hydraulic Cylinder 64
5.3.3 Induction Electric Motor 66
5.3.4 A Simple System 68
5.4 Power Flow Directions, and Causality 70
5.4.1 Introduction 70
5.4.2 Directions of Power F l o w 7O
5.4.3 Causality 71
5.4.4 Exa/m~le: Hydraulic Cylinder 76
5.4.5 Example : Induction Electric Motor 76
5.4.6 Example: Simple cylinder-Load System 78
5.4.7 Summary 78
5.5 Preparing Equation Set 81
5.5.1 In troduction 81
5.5.2 Example: A Simple Component 81
5.5.3 An Elementary System Example 83
5.5.4 A More Cc~plete System 84
5.5.5 Summary 88
5.6 Some Further Aspects of Bond Graphs 89
5.6.1 Modulation of Effects 9O
5.6.2 Modulated Transformers 9O
5.6.3 Fields and Junction Structures 92
5.6.4 Simplifications 92
5.7 Conclusion 93
5.8 References 93
CHAPTER 6 SOLUTION OF POWER FLOW MODELS 94

6 .i Introduction 94
6.2 Digital Simulation 95
6.3 Expression-Orientated CSSL' s 97
6.4 Conclusion 98
6.5 References 99
CHAPTER 7 SELECTING EQUATIONS AND COEFFICIENTS i01

7.1 Introduction 101
7.2 Compliance 102
7.2.2 Oil Compliance 102
7.2.3 Values for Bulk Modulus 106
7.2.4 Mechanical Compliance 106
7.3 Friction 109
7.4 Modelling Driven Loads 110
7.4.1 Inherent Load s ii0
7.4.2 External Load Forces 112
7.5 Leakage Flowrate 112
7.5.1 Re laticn shlps 112
7.5.2 Coefficients 114
7.6 Relief Valve Flowrates 116
7.6.1 Relationships 116
7.7 Electric Induction Motor 118
7.7.1 Re lation ships 118
7.7.2 Model 120
7.7.4 Conclusion 121
7.8 Hydraulic Pumps 121
7.8.1 Relationships 121
VI
CHAPTER 7 (contd.)
7.9 4-Way Control Valves 125
7.9.2 The Basic Relationship 126
7.9.3 Valve Flow Nomenclature 129
7.9.4 The Closed-Centre Control Valve 129
7.9.5 The Open-Centre Control Valve 132
7.9.6 The Tandem-Centre Control Valve 136
7.9.7 Summary 138
7.10 Actuators 139
7.10.2 Linear Actuator 139
7.10.3 Rotary Actuator (Hydraulic Motor) 142
7.10.4 Equations and Coefficients 142
7.11 Some Other Common Components 144
7.11.1 Hydraulic Lines 144
7.11.2 Filters 146
7.11.3 Accumulators 146
7.11.4 Some Other Valves (Check Valve, 148
Counterbalance Valve)
7.12 Conclusion 151
7.13 References 152
CHAPTER 8 APPLICATIONS OF BOND GRAPHS 153

8.2 Closed-Centre Valve-Controlled Inertia Load 153
with Friction
8.3 Loaded Hydraulic Servosystem 155
8.4 Pump Sub-System 156
8.5 Pump-Controlled Hydrostatic Drive 164
8.6 Valve-Controlled Hydrostatic Drive 166
8.7 A Lifting System 169
8.7.2 Development of Bond Graph 172
8.7.3 The Equations and Coefficients 173
8.7.4 Simulation 173
8.8 A Highly Dynamic Electrohydraulic Control System 173
8.8.2 The Bond Graph 178
8.9 Conclusion 181
8.10 References 182
CHAPTER 9 OPTIMIZING DYNAMIC RESPONSE 183

9.2 The Requirements 184
9.3 Error Criteria 185
9.4 Search Procedure 188
9.4.2 Single-Parameter Optimization 188
9.4.3 M u l t i - P a r ~ e t e r Optimization 189
9.4.4 Complex 190
9.5 Example 192
9.6 Conclusion 192
9.7 References 194
CHAPTER i0 PHENOMENA WHICH CAN AFFECT RESPONSE 196

Vil
CHAPTER i0 (contd.)
10.2 cavitation 197
i0.2.1 General Discussion 197
10.2.2 Cavitation in Modelling and Simulation 200
10.3 Hydraulic Backlash 201
10.4 F l o w Forces in V a l v e s 204
i0.5 Hydraulic L o c k 205
i0.6 Contaminated Fluid 210
i0.7 Conclusion 212
10.8 References 212
C H A P T E R ii CONCLUSION 215
APPENDIX 1 STATIC DESIGN APPROACHES 217
APPENDIX 2 SI CONVERSION FACTORS 221
APPENDIX 3 DYNAMIC RESPONSE - - A BIBLIOGRAPHY 222
INDEX 226
CHAPTER 1
INTRODUCTION
i.i INTRODUCTION
Hydraulic control systems are used to control the position or speed of resisting
loads. Final drive is usually provided by a hydraulic actuator either a linear-
motion hydraulic cylinder or a rotary-motion hydraulic motor. The actuator develops
its force or torque by receiving liquid from a positive displacement pump at a
relatively high pressure, usually between 7 and 35 MPa. The actuator develops its
motion by receiving flowrate of liquid from the pumping system. Fig. i.i illustrates
several simple hydraulic control systems.
Hydraulic control systems are used
where relatively large forces or torques are required (industrial presses;

mobile lifting, digging, and materials handling equipment, etc.),
where fast, stiff response of resisting loads is required (machine tool

drives, flight simulators, etc.),
where close control of response is required (aircraft control surfaces,

machine tool slides, industrial robots, etc.),
where manual control over substantial powered motions is required (mobile

equipment, aircraft controls, road vehicle steering, etc.),
as the final actuator sub-system in complex automatically controlled

situations (electrohydraulic flight simulators, industrial robots, fatigue
and other programmable test rigs, etc.).
2
a. A Linear Drive Hydraulic Control System

primaryDrive
HydraulicPump
~.~ --- Hydraulic Accumulator
I I HydraulicCylinder7
i .
~-----~ilter I ~ InertiaLad~
I~ ~ ~ %---4-WayManualControlValve
II ~-Pressure
ReliefValve
I] II Reservoir
b. A Rotary Drive Hydraulic Control System

HydraulicMotor
~NN
f
i:
FIG. i.i TYPICALHYDRAULIC CONTROL SYSTEMS

That is, they are used in dynamic situations - - situations where response to commands
can't be instantaneous, but must be controlled and very often optimized to follow
pre-determined motion-time paths. Fig. 1.2 illustrates two simple command and
response pairs.
It is not sufficient for the designer of a hydraulic system to know that his
proposed system WiZ~ move the driven load from one state to another. He should know
also how the load will move. He needs to appreciate not only the initial and final
states of the responses, but also the time-domain path between these states (Fig. 1.2).
He should know if system response is stable, if it is fast enough or too fast, if it
is oscillatory or aperiodic, etc.
Dynamic analysis can be
corrective in nature, the objective being to find out why an existing system
is not performing satisfactorily and to deduce what can be done to improve the
situation,
synthetical in nature, with the objective of ensuring good dynamic response

before commitment to system hardware.
In the latter situation, the proposed system's dynamics are designed as part of the
system design process. The present Text is mostly concerned with this approach,
though it is relevant also to corrective analyses.
1.2 A DESIGN PHILOSOPHY
Fig. 1.3 illustrates the stages in a dynamic design philosophy. Specify Task
does not imply use of any particular system or technology - - simply a definition of
the task to be accomplished. Define Required Response is also a consideration
independent of the particular system or technology to be employed ....... typically this
would be a time-based curve of the desired system response form, with time and response
magnitudes as explicit scales, such as is shown on Fig. 1.2.
The stage Propose System requires the first corm~itment to a technology in

the present Text a decision to use a hydraulic control system, whose form is then
proposed. Initial selection of the hardware is decided usually by a combination of
past experience,
the maximum pressure to be allowed,
static power and overall motion considerations,
the available hardware components, and their cost.
a. Linear Drive b. Rotary Drive
Control Valve Motion Desired Speed, rpm
Full Opez 5OO
200~
time, s I time, s
Close(
!
0 0.5 ~v 0 0.5
Load Motion System Speed, rpm
i m. 50~
0.5m-
me, s . . . . . . . Re; s
0 200
0 0.5 1 0 0.5
FIG. 1.2 SOM~ COMMAND-AND-RESPONSE PAIRS

(Response times can be much faster or
slower than those shown, depending on
system and load and input)
It is after this initial decision that a more conclusive and enlightening dynamic
response analysis should be made, requiring the steps Develop D y ~ i c Mode~ and Prsdiot
Respon88 (by solving the model) . The predicted response is compared with the desired
response (Compare) ana the discrepancy between them is assessed (Judge Response
A decision is made then that the performance of the proposed system is
adequate (Response Acceptable; Finalize Design),
encouraging enough so that simple parameter adjustments will make it

acceptable (Response Promising; Optimize It),
relatively hopeless, requiring re-design of the proposed system (Response

Hopeless; Re-design System).
Dynamic performance analyses are often carried out for the high-cost hydraulic
systems associated with aircraft, military, space, and some machine tool applications.
They are less often associated with the design of industrial systems. One result is
that while many such systems are accepted by the user, their performance rarely
approaches a considered optimum.
Relatively few mechanical (or other') engineers are confident and competent with
predictive dynamic analysis of systems. Predictive dynamic analysis hinges on the
formation of a mathematical representation of the system. ~is mathematical model
is used to extract information on the response characteristics of the system. If an
adequate mathematical model can't be formed quickly and with confidence, dynamic
analysis is not an attractive proposition to most designer-analysts.
Few designers of hydraulic control systems are able to confidently form dynamic
models of proposed systems. On the other hand, given a mathematical model, most
engineers can arrange for its solution, which provides the required prediction of
dynamic performance. Generally, solution of anything other than the simplest models
is best effected by digital computer. Digital simulation procedures, widely
available in easy-to-use forms, are of particular relevance. There is no need for
the designer-analyst to be himself a computer programmer. Given the means to solve
(simulate) his model, the designer-analyst can change one or more of the model's
parameters in order to seek a desired form of system response. Such optimization of
response can be sought through repetitive solutions of the model as parameters are
changed. Automated optimization procedures can also be used.
Accepting that solution of mathematical models is relatively easy, the great

hurdle for the designer-analyst lies in the formation, with speed and confidence, of
i. Specify
Task
2. Define Required Response,

For Example
~'r-t
0 ~ t2 tr
Input Expected Response R e q u i r e d
~ 3. Propose System, I
and its Hardware, I
to Accomplish 1
Task I
4- Develop Dynamic
Model
(Solve Model) " ~I 6. Compare I
I l
~ ~esponso L 7. Judge Response I J9 Promising;
Response I
I Hopeless.
Redesign System I TM
Quality; Decide 'i
Between 8, 9, I0 vI Optimize it
Acceptable.
Finalize Design
FIG. 1.3 APPROACH FOR DESIGNING DYh~AM_IC RESPONSE

7
an adequate mathematical model of his proposed system.
1.3 DYNAMIC MODELLING
Conventional dynamic modelling procedures for control systems, such as
transfer functions,
block diagrams and signal flow graphs,
sets of equations not having a unique formal arrangement,
vector-matrix (state space) expressions,
have been around for a long time. Journals contain many examples of successful
applications of each of them. Yet the lack of general use of these model forms by
designers of hydraulic systems suggests that they are insufficiently attractive.
It appears that the conventional dynamic modelling procedures do not adequately

provide for sc~e of the designer's needs, including
the need for easy-to-follow formality in the modelling process,
the need for reliable re-useable models of common hydraulic system components.
Hydraulic control systems are of modular construction, being made up by
assembling and interconnecting a set of well-defined off-the-shelf hardware
devices. A particular type and size of pump, valve, actuator, etc. is used
in many different systems. The designer needs to have available a set of
dynamic models, one for each common hydraulic system component, which is
capable of being interconnected to form a system model in a manner reflecting
the formation of the hardware system itself,
the need to be able to change easily the model of a proposed system to reflect
conceivable variations in the proposed system, aimed at ensuring choice of the
best design for the particular system.
An engineering designer needs to be creative - - to have imagination ........for

. the
conception of design solutions. For subsequent analyses however, he requires
availability of well-established and readily-applied techniques.
Hydraulic control systems have some distinctive features relating to dynamic

analysis, including the following:
They are usually of relatively small scale. In terms of a transfer function,

their dynamics often can be described adequately by a second, ~*ird, or fourth
order differential equation. Rarely do they require more than sixth order
transfer functions, although complex multi-axis systems occur. Also, they
are often compact and amenable to a lumped parameter approach. Thus, they
are potentially amenable to dynamic modelling;
They are inherently non-linear due to the presence of turbulent orifice flow,
friction, valve lap, pressure limiting actions, and other effects;
They usually are expected to provide large scale response. Quite normally
they are commanded to move the controlled load from one extreme position to
the other;
These latter two factors cause linear analysis techniques to be of very

limited use;
They are power transducing, power transmitting systems. Typically, power is

drawn from a source such as an electric mains supply, diesel engine, aircraft
engine, etc., and progressively converted and transmitted to a mechanical
form useable at the driven load.
Conventional modelling procedures do not particularly recognize these factors, nor

particularly utilize them to advantage. The reason-for-being of this Text is the
proposition that dynamic power exchanges taking place between system components provide
the key for incorporating dynamic response analysis into the system design environment.
POWER FLOW MODELLING procedures allow a dynamic description of the system which retains
a close association with the physical connections and actions of the modular system
itself.
Power flow modelling, and in particular its recent extension POWER BOND GRAPHS,
provides the designer of powered control systems with a dynamic modelling procedure
which is more fundamental to, and more useable in, the system design environment.
It is intended to demonstrate this by the end of the Text. It is intended also to
leave the reader with a library of power bond graph models of the more common hydraulic
system components, and with the ability to develop power bond graph models of any
additional components he might wish to use.
Fig. 1.4 (a) illustrates a simple hydraulic system, Fig- 1.4 (b) a power bond graph
structure for the system, and Fig. 1.4 (c) the set of equations which is a dynamic
model of the system. The bond graph structure is the key for linking the system to
its dynamic model. The bond graph is a type of signal flow diagram which to a person
familiar with bond graphs bears a striking affiliation with the hardware system
diagram. The set of equations is readily formed from the bond graph structure.
Each line in the horizontal chain of Fig. 1.4(b) represents a physical component
through which power flows between the adjacent system components. The product of
a. The S y s tern
A ....... Fa Ira
r
I / /,t,,,,,ILr
............ ~
I Friction
b. Power Bond Graph

(Return flow n e g l e c t e d , Xv positive only)
Rv Rf
....... ~ "~'1 ~-'~ 0 ~ T F ...... Fa ' "~11

ps ' Qs Qa A Xm
Ca Im
C. Dynamic Equations
Ps = constant (assumption)
Qs = K - Xi - ~Pv ~ (for Xv positive)
Pa = Pa(o) + 1~Ca IQc dt (linear c a p a c i t a n c e equation)

Ff = R(Xa) = friction force for c y l i n d e r
Fm = Fa- Ff
e
= X(O) + 1/Ira . IFra d t (Newton's law)

Fa = Pa- A
Qa = l~-a/N
APv = Pv- Pa
Qc = Os - Q a
FIG. 1.4 AN ELEMENTARY HYDRAULIC CONTROL SYSTEM

AND ITS BOND GRAPH MODEL
10
the two variables written against each line describes the power flowing in the line.
Thus Ps - Qs (supply pressure times supply flowrate) describes the power flowing
into the control valve. Similarly, Fa - Xm (force applied to the mass times
velocity of the mass) describes the power flowing into the driven load. The R ,
C , and I symbols appearing at the ends of the spur lines of the diagram indicate
respectively the presence of resistive (R) , capacitive (C) , and inertive (I)
effects which affect the system's dynamic response. The half-arrows on each line
indicate the power flow direction in each power bond, either real or assumed. The
short transverse bar at one end of each line indicates a cause-effect decision aimed
at arranging equations, to be formed later from the bond graph structure, in a form
most suitable for simulation. These various points will be developed and explained
in the Text. The Text's nomenclature and units are discussed in Chapter 2.
Formation of the model structure, i.e. of the power bond graph, decisions on the
precise relationships to be used in the equations, and formation of and arrangement
of the equation set, are separate and progressive steps in the dynamic model derivation
procedure. In conventional modelling procedures these steps become jumbled together,
often creating confusion in the minds of part-time dynamic analysts. It will be
demonstrated in later chapters that power flow models allow the formality and
flexibility required in the system design situation, particularly in that
component models are useable in a variety of system models, as are the hardware
components themselves useable in a wide variety of systems,
system models are readily formed by direct coupling of component models, as

are systems formed from the components,
local changes in model structures are easy to make,
model relationships can be varied locally without disrupting the remainder of

the model.
Power flow and power bond graph modelling procedures can be used with all types
of powered control s y s t e m s - - e l e c t r i c a l , pneumatic, mechanical, etc. The present
Text however concentrates on their use with hydraulic systems, although this
encompasses electric motors, electrohydraulic devices, and a variety of mechanical
sub-systems. The ideas of power flow and power bond graph modelling have been
developed by others, and references will be provided in appropriate places. The
present Text simply utilizes power flow techniques and orientates them specifically
towards use in the design and analysis of hydraulic control systems.
Beyond a review in Appendix i, with References, the present Text does not provide
instruction on how to develop a basic design for a hydraulic system for a specific
I!
application. It concentrates on the inclusion of dynamic response analysis as part

of the system design procedure.
1.4 DESIRABLE BACKGROUND
The user-reader of the Text ideally would have as background
familiarity with basic hydraulic control system components, systems, and the
symbols used to represent hydraulic components in system diagrams of the type
of Fig. i.i,
familiarity with hydraulic system design to the extent of being able to

select for specified applications the major components such as pumps, relief
valves, control valvesr and actuator st
familiarity with engineering dynamics to at least the stage of appreciating

common physical concepts and relationships such as Newton's law, the
potential drop-flowrate-resistance concept, the potential-flowrate-
capacitance or compliance concept, etc. It is basic to all dynamic
modelling procedures that such actions and relationships be appreciated,
familiarity with some of the conventional modelling procedures used in

first-course control theory or dynamic systems analysis.
The Text is primarily aimed
at practising designers of hydraulic control systems, and engineers who from

time to time must design or specify the performance required of hydraulic
systems,
at persons interested in control engineering, automation, or design whose

duties or vision encompasses possible use of hydraulic control systems,
at students in specialized undergraduate courses, graduate courses, and

extension courses, involving dynamic analysis and fluid power systems.
1.5 ARRANGEMENT OF TEXT
The Text is arranged as follows:
Chapter 1 provides the setting for the Text and makes its case;
Chapter 2 provides a rational Nomenclature which aids in easy recognition of

symbols in equations. It also discusses the use of SI units and provides
12
a table of those used in the Text;
Chapter 3 provides a review with examples of conventional modelling

procedures used in dynamic analysis~
Chapter 4 introduces and demonstrates the power flow modelling approach,

providing the basis for developing power bond graph techniques;
Chapter 5 introduces power bond graphs, the key feature of the Text, which
is offered as a major step towards making predictive dynamic analysis viable
for more designers of hydraulic control systems;
Chapter 6 reviews briefly digital simulation of power flow models, taking

the view that this is the most practical and most available approach to the
solving of models;
Chapter 7 discusses and demonstrates the choice of equation and relationship

forms for physical actions and devices. It also discusses choice of
numerical values for coefficients in the equations;
Chapter 8 describes some applications of bond graph modelling and digital

simulation of hydraulic systems;
Chapter 9 briefly outlines response optimization techniques relevant to

hydraulic system design;
Chapter I0 briefly introduces some factors which can affect the performance
of hydraulic systems, and about which the system designer should be aware;
Chapter ll is a summary and conclusion;
Appendix 1 reviews briefly conventional (static) design approaches for

specifying the components of hydraulic systems and their size ratings;
Appendix 2 provides conversion factors for British to SI units, for the

quantities used in the Text;
Appendix 3 provides a limited bibliography on dynamic analysis of hydraulic

control systems.
1.6 CONCLUSION
In conclusion, the orientation attempted with the Text can be stated: "Given a
proposed hydraulic control system, simulate it to check and to optimize its dynamic
performance before any final commitment is made to expensive hardware".
CHAPTER 2
NOTATION AND UNITS
2.1 INTRODUCTION
It is difficult enough to master reported dynamic analyses without the problem

of inconsistent and non-standard notation and units. It is inevitable that many
subscripted symbols will be used, which adds to the confusion. Various authors use
British, metric, and Syst~mes Internationales (Sl) units, and sometimes mix dimensions
within their chosen unit schemes.
What a help if everyone would use the same notation and units' This would allow
immediate reCognition of symbols, and of the significance of the magnitudes of
coefficients and numbers appearing in relationships. The value of this kind of
recognition in the system design environment should not be under-estimated. The
designer needs to be able to quickly utilize established techniques and data, rather
than seek out new unproven and possibly ambiguous materlal.
The notation, the way of writing relationships, and the units used in this Text
seek standardization.
2.2 NOTATION
2.2 .I Symbols
Fig. 2.1 lists the symbols used. General criteria are.
Upper case English letters are used generally for system variables and
coefficients, each being reserved for particular quantities; some widely-
14
a. Parameters (Coefficients and Variables)
Capitals Lower Case Greek
A -- a r e a a - - angular acceleration
B - bulk modulus of oil b - B-
C - capacitance, or capacitive c - y -
effect, or compliance d- -
D - operator d/dt e - E -
E - general symbol for effort f- - damping ratio
(i.e. potential) variable; g - gravitational - efficiency
also voltage acceleration 8 - angular displacement
F - force h - I
G - i - electrical K
GY - gyration power transformation current
H - j- - coefficient of
I - inertive or inductive effect; k - friction
~ -
inertia or inductance v - kinematic viscosity
J - m -
K - general symbol for coefficient n - 0 -

L - length o -
-
LR - lever ratio p - p - density
M - q - -
N - revolutions per unit time r - T - time constant
O - s -
u -
P - pressure t - time
Q - volumetric flowrate; general u -
X -
flowrate variable v -
R - resistance, or resistive w - - angular velocity
effect x -
S - source; constant or y-
independent variable Z -
SE - source of effort (potential)
SQ - source of flowrate
T - torque
TF - direct power transformation
U -
V - volume
W -
X - linear displacement
y -
Z -
b. Suffices (all lower case)
a - actuator (cylinder or motor) n -

b- filter, strainer o - oil
c- fluid compliance (compressibility) p - pump
d - q -
e - exhaust (tank, or return line) r - relief valve
f - friction s - supply
g - gravity t-
h - hose, or fluid line u- coulomb friction
i - input v- valve
j - rotating inertia w -
k - spring, mechanical compliance x -
- leakage y-
m - linear motion inertia (mass) z -
i, 2 ... 9 denote different local values
FIG. 2.1 STANDARD NOTATION

15
u s e d Greek l e t t e r s are retained; for g r a v i t a t i o n a l acceleration, for time,

and for e l e c t r i c a l current, g , t and i r e s p e c t i v e l y are retained;
S u b s c r i p t s are lower case E n g l i s h letters, each r e s e r v e d for a p a r t i c u l a r

significance; n u m b e r s are u s e d also;
C u r r e n t l y a c c e p t e d u s a g e and p h y s i c a l a s s o c i a t i o n o f symbols w i t h p a r t i c u l a r
q u a n t i t i e s are r e t a i n e d as far as is feasible;
S u b s c r i p t s are w r i t t e n in ~ i ~ w i t h their p a r e n t symbol; m u l t i p l e subscripts

are allowed;
It is a s s u m e d that the reader will k n o w o r w i l l q u i c k l y realize w h i c h

p a r a m e t e r s are d y n a m i c (time dependent) and w h i c h are constants. The
symbology (t) to m a k e the d i s t i n c t i o n is r a r e l y used, e.g. P and not
P (t) is u s e d for a d y n a m i c pressure. If a n o r m a l l y d y n a m i c p a r a m e t e r is
a s s u m e d to b e constant, it is p r e f i x e d w i t h S to denote Source, e.g. SP
d e n o t e s a c o n s t a n t pressure.
2.2.2 E x a m p l e s o f Symbol Use
Ck denotes compliance (C) of a spring (k) ;
Bo d e n o t e s the b u l k m o d u l u s (B) o f oil (o) ;
Ip d e n o t e s inertance (1) of p u m p (p) rotating parts;
Rip denotes a resistance (R) expression, o r a resistance coefficient,

d e s c r i b i n g leakage () from a p u m p (p) ;
Kvl denotes a particular (i) valve (v) coefficient K , w h e r e there is

m o r e than one such c o e f f i c i e n t p r e s e n t ( Kv2 , etc.);
8a d e n o t e s the a n g u l a r d i s p l a c e m e n t o f a h y d r a u l i c m o t o r (actuator).
2.2.3 Operations
M u l t i p l i c a t i o n is i n d i c a t e d b y a central dot: N P ; Np Ck ; 13 Xm ;
Decimal p o i n t is a low dot: 9.8 ; 9.8 - Np ;
D i v i s i o n is indicated b y slash: N/P ; Np/Ck ; 9.8/Np ;
B r a c k e t s are u s e d only w h e r e n e c e s s a r y to p r e v e n t ambiguity: Ac/(Vc Bo) ;
Exponents appear in a normal raised position, p l a c e d i m m e d i a t e l y a f t e r the

l a s t s u b s c r i p t and r a i s e d just above it: X2 ; Xm 2 ; Xm3 2 ; ~n 0"5 ;
18
Integration is indicated by the usual
before dt : IPa dt ; ~Pa C2 dt ;

I ...dt , with one space, but no dot,
J J
The overdot syst~n is used for time derivatives; linear velocity is X ;

linear acceleration is X , etc.;
Raising ten to a power is expressed i0 Ex or i0 E-x ( i0 x and i0 -x

respectively).
2.2.4 Equations
The following are typical expressions appearing in this Text:
QEp = Pp/REp
Qp = Kv Xi dp
= ~m(0} + z / m - I~ dt
BO = 1750 E6 (i.e. 1,750 x 106 ) Pascal
2.2.5 Comments
A little usage will show that this scheme is easy to write, easy to type,
easy to understand, and is easily associated with digital computer programming
and print out.
Some of the terms included in Fig. 2.1 apply specifically to power bond graph
terminology to be introduced later.
Only symbols used in this Text are designated in Fig. 2.1. There are plenty
of gaps for future assignment or general use.
2.3 UNITS
Syst~mes Internationales (SI) units are used without exception. Apart from its
substantial and growing international acceptance, the SI system overcomes most of the
ambiguity associated with the use in engineering and science of the British and the
standard metric systems of units.
The SI units which appear in this Text are shown in Fig. 2.2. Appendix 2
provides some useful conversion data.

17
Conversational
Physical Quantity Name of Unit Symbol of Unit
Alternatives
Basic Units
Length metre m centimetre (cm) ,
millimetre (mm)
Mass kilogram kg tonne (E3 kg)
Time second s minute s
Electric current ampere A
Temperature degree celsius C
Derived Units
Area metre 2 m2
Volume metre 3 m3 litre (E-3 m 3)
Density kilogram/me tre 3 kg/m 3
Force (also weight) newton N kilonewton (kN)
Torque newton metre N - m
Rotating inertia kilogram metre 2 kg - m 2
Pressure pascal (N/m 2) Pa kPa, MPa, bar(E5 Pa)
Energy (work) joule J
Power watt W kilowatt (kW)
Electric potential volts E
(electric voltage)
Electric resistance ohm
Velocity metre/second m/s
Acceleration metre/second 2 m/s 2
Flowrate metre 3/second m3/s litre/sec or
litre/minute
Angle radian rad revolution
Angular velocity radian/second rad/s revolution/sec or
minute
Frequency hertz Hz cycle per second
Dynamic viscosity newton second/metre 2 N s/m 2, centipoise (cP)
or pascal second or Pa o s
Kinematic viscosity metre2/second m2/s centistoke (cSt)
Common multiples: micro (u) E-6 ; milli (m) E-3 ; centi (c) E-2 ;
kilo (k) E3 ; mega (M) E6
FIG. 2.2 RELEVANT SI UNITS AND SYMBOLS

(See Appendix 2 for Conversion Factors)
18
2.3 .I Comments
In c o n v e r s a t i o n o r in d e s c r i p t i v e passages, convenient multiples of b a s i c SI
units can b e useful. For example, flowrate in l i t r e s / s e c o n d (/s) ;
pressure in m e g a p a s c a l s (MPa) ; force in k i l o n e w t o n s (kN) ; angular
velocity in revolution per minute (rpm) . The author considers the p r e s s u r e
u n i t bG/ to be an u n n e c e s s a r y bastardization o f the SI s y s t e m a n d d i s c o u r a g e s
its use (i bar = 105 Pa) .
For numerical work (specifying coefficients, applying relationships, etc.),

use only base SI units, for exa~[ole, m3/s for flowrate, Pa for pressure,
N for force, and r a d i a n / s e c o n d (rad/s) for a n g u l a r velocity. Failure to
do this can lead to gross and u n s u s p e c t e d magnitude errors in c a l c u l a t i o n s ,
with I0 E -+ n , n = 1 ... ~ , floating around. One can e a s i l y enough talk
or w r i t e descriptively of i0 M P a , b u t should w r i t e it as i0 E6 P a w h e n it
comes to equations.
For unit products use central d o t to s e p a r a t e component units
i.e. N- s , not Ns , for N e w t o n second,
kg -m 2 , not k g m 2 , for k i l o g r a m m e t r e 2.
Take care w i t h u n i t e x p o n e n t s where slash is u s e d
i.e. m 2 - P a -I - s -I = m2/(pa s) ~ m2/pa s
2.4 REFERENCES
For a m o r e complete description o f the S.I. s y s t e m and its units see:
2.1 Blackman, D.R., SI units in Engineering, MacMillan, 1969.
2.2 Chiswell, B. and Grigg, E.C.M., SI units, J. Wiley, 1971.
2.3 The I n t e r n a t i o n a l S y s t e m o f Units, Nat. Bur. Stand. (U.S.), special

p u b l i c a t i o n No. 330, 1972.
2.4 The A B C ' s o f SI, National Fluid Power Association, USA, 1974.
CHAPTER 3
CONVENTIONAL MODELLING PROCEDURES
3.1 INTRODUCTION
Transfer functions,
block diagrams and signal flow graphs,
vector-matrix (state-space) equations,
informally arranged sets of dynamic and algebraic equation,
are regarded in this Text as conventional forms for dynamic models. They are
conventional in that each is widely used in hydraulic system analyses readily
available in texts and articles.
Readers familiar with the application of conventional modelling procedures to

hydraulic control systems can pass over this chapter. Others may prefer to delve
into it to whatever degree is needed for revision or orientation. The Chapter
assumes familiarity with the above model forms, and simply demonstrates briefly their
application as a prelude to development of power flow and power bond graph modelling
in following chapters. The distinctions and advantages of the power flow procedures
will be developed in the later chapters.
3.2 CONVENTIONAL MODEL FORMS
Fig. 3.1 a illustrates an inertia load Im with frictional damping Rf and

support compliance Ck , driven by a hydraulic ram (Area A ) which is controlled by
a 4-way closed-centre valve (volumetric flow coefficient Kvl , pressure flow
coefficient Kv2 ). Load position feedback is provided through the mechanical
linkage L1 , L2 , the assembly thus becoming a servomechanism with lever position
20
a. System Schematic
Xi
supply Oil
const, press, ps
XV
.......... , ......
Kvl, Kv2--
\\ L2
Xm
Rf
b. Equation Set Model
xv = (L2 - L I ) / L 2 Xi - L I / L 2 Xm 3.1
Qs = Kvl " Xv - Kv2 - Pa 3.2
Qs = A - ~m+v/B ~a 3.3
Im ~m = Pa " A - Rf - Xm - Ck - Xm 3.4
c. Transfer Function
X m = K1
Xi K2 D3 + K3 D2 K4 D + 1
where K1 = (L2 - L I ) / [ L I 2 / L 2 + L1 K V 2 C k / ( K v l A)]

K2 = V " I m / [ B K v l A L I / L 2 + B Kv2 Ck]
K3 = [V " R f / B + Kv2 I m ] / [ L I / L 2 K v l A + K v 2 Ck]
K4 = [A 2 + V * C k / B + K v 2 " R f ] / [ K v l A L I / L 2 + K v 2 Ck]
d. Vector-Matrix Model, of form X =A X + B u
= -Ck/Im -Rf/Im A~Im I~I + 0 Ixil

a - L I / L 2 Kvl. B / V -A. B / V - K v 2 B / V a (L2 - L I )/L2 "Kvl B / V
e. Block Diagram
I D ...........
f. Signal Flow Graph
B~~/(V-D) -A
Xi'-- ( L 2 - L I I / L 2 . Kvl ~ __
-LI/L2
FIG. 3.1 SOME CONVENTIONAL MODEL FORMS FOR SYSTEM OF FIG. 3.1a
21
Xi as the controlling input and load position Xm as the controlled output. Supply
pressure Ps is assumed to remain constant. The lever system L1 , L2 is such as
to allow small control valve motions Xv and much larger load motions Xm . Xi is
a m a l l motion. This system will be used to demonstrate conventional model forms.
3.2.1 Informal Equation Set
A dynamic model may be in the form of a set of simultaneous equations, linear

and non-linear, algebraic and differential, describing the relationships between the
system variables and coefficients and the inputs. The arrangement of the equations
is chosen by the analyst. For example, the dynamic response of the system of
Fig. 3.1 a can be described by the equation set of Fig. 3.1 b , in which Qs
describes the flowrate of oil following input Xi ; Pa describes the pressure
developed in the cylinder due to Qs ; B is the bulk modulus of the oil; V is the
average volume of oil under compression in the cylinder ( V is more accurately
expressed as V(O) + A Xm , which allows for the expanding cylinder volume.
However, this introduces the non-linear product Xm Pa ); and the other parameters
were previously described.
Depending on the analyst the equations 3.1 to 3.4 could be differently arranged,
or even different in content. The present equations are all linear, whereas several
non-linearities can readily be included if the analyst considers them to be important
to his response predictions.
3.2.2 Transfer Function
This is the most common form of model. A transfer function is a linear

differential equation with constant coefficients, in which a chosen response variable
is related to a specific input variable. Fig. 3.1 c shows the operational way such
a differential equation is expressed, in which D denotes the operator d/dt .
simple cross-multiplication will yield the conventional differential equation. The
transfer function can be derived by combining equations 3.1 to 3.4 so as to eliminate
all variables other than input Xi and response Xm .
The transfer function is particularly descriptive, as it describes the time

dependent ratio of response to input and is independent of input form, though not of
the input variable chosen. Analysis and model solution procedures based on transfer
functions have been highly developed (stability criteria, root locus, frequency
response, system compensation, equation solutions, etc.). However, the necessity of
linearization reduces their usefulness for hydraulic control system analyses. The
coefficients can be awkward combinations of system parameters ........-..see the definitions
of K1 to K4 on Fig. 3.1 c . Such awkward arrangements are not helpful in the
system design situation.
22
3.2.3 Vector-Matrix Models
Vector-matrix techniques allow concise representation of the dynamic relationships

between a number of system variables and a number of simultaneous inputs. Fig. 3.1 d
shows the vector-matrix model of the system under discussion. The chosen state
variables are Xm , ~ and Pa . Only one input, Xi , is included: however the
model is readily extended to include other simultaneous inputs to the system
possibly an external force Fi applied to the inertia in the line of motion. In
essence, the system's state variables are related to the inputs experienced by
inherent system parameters which make up the A and B matrix entries.
In effect, the vector-matrix model describes concisely a set of simultaneous

transfer functions each of which relates a specific system response variable (state
variable) to a specific input variable. While vector-matrix equations can include
non-linear entries, most subsequent formal vector-matrix analyses apply only to
linear situations.
3.2.4 Block Diagrams and Signal Flow Graphs
Both are diagrammatic representations of a system's dynamic equations.

Figs. 3.1 e and f show the block diagram and the signal flow graph respectively
for the system of Fig. 3.1 a . The diagrams can be formed from the system equations.
However, many analysts prefer to form the block or signal flow diagram direct from
the system schematic, thinking the system dynamics out as they construct the diagram.
A transfer function or vector-matrix or equation set model is readily deduced from a
completed block diagram or signal flow graph. Alternatively, the diagram itself is
readily simulated for response solution. There is no particular advantage of block
diagram relative to signal flow graph, or vice versa. The latter is a little more
economical of linework, but the former bears a closer equivalence to the physical
system. Both are capable of ready reduction (if linear) to a transfer function or
vector matrix model, though this is hardly necessary or even desirable in this day of
computers. Non-linear terms can be included on block diagrams and signal flow graphs,
but care must then be taken in manipulating the diagram.
3.3 DERIVING THE MODEL
All approaches to modelling require insight of the physical laws and relationships
which govern the b e h a v i o u r o f the various devices used in the system components,
together with an appreciation of the interaction of the components and of the structure
of the system. The successful formation of dynamic models is an art as much as a
science, and experience and experiment are important assets.
The procedure for developing a model can be summarized:

23
i. Obtain a diagrammatic representation of the system to be modelled, showing

clearly h ~ the system is intended to function;
2. Select that response variable (or set of variables) study of which will
provide clearest appreciation of system response;
3. Designate an input (or set of inputs) which will cause the system to respond
in a realistic operational mode. It is not necessary to specify the form
of the input at this stage, merely its physical variable and location;
4. List ass~nptions which are valid in the particular circumstances, and which
will render modelling easier but still adequately valid. Typical
assumptions which might be valid include: input drive speed constant;
pressure relief valve does not lift; all driven inertias can be lumped in
with the designated load inertia; all friction is viscous; pressure in the
return lines is negligibly small (zero); motor-load coupling is rigid; the
model is for small scale response predictions only; etc. This is an area
requiring considerable thought for each modelling attempt. However, there
is a lot of confirmed past experience available in the literature;
5. Imagine the system to be in equilibrium. This can be at zero value state,

or at some preferred steady-state operational state. Note all initial
values of system variables, if other than zero;
6. Imagine the designated input to be applied at the instant designated as

t =0 ;
7. Form equations describing the subsequent behaviour or actions of the various

physical components which comprise the system.
The relationships so formed can be expressed in equation or signal flow diagram

forms as previously discussed. It is necessary to ensure
that correct relationships are used;

that all relevant relationships are included.
It is desirable that dynamically irrelevant relationships be omitted in order to have

the simplest model adequate for the particular analysis to be made.
The equations should be algebraic rather than numerical. The substitution of

numbers for algebraic representation of parameters is best left until the model has
been developed and is ready for solution.
24
3.4 S E L E C T I O N OF R E L A T I O N S H I P S
F o r a p a r t i c u l a r model, some of the r e l a t i o n s h i p s to b e used will b e clear cut.

For example, the origins of e q u a t i o n s 3.1 (lever ratios) and 3.4 (Newton's law) should
b e obvious. A few r e l a t i o n s h i p s m i g h t b e m o r e ambiguous, r e q u i r i n g j u d g e m e n t b y the
analyst. E q u a t i o n 3.2 is in this category. E q u a t i o n 3.2 is a w e l l k n o w n l i n e a r
e x p a n s i o n of the p h y s i c a l l y obvious e x p r e s s i o n for flow through the valve,
Q = f(Xv, (Ps - Pa)~ 3.5
w h i c h is o f t e n good for small-scale response (perturbation) analyses. However, if

large-scale response (large Xm ) is required, the a n a l y s t m a y p r e f e r the well k n o w n
orifice flow form,
Q = Kv X v (Ps - Pa) 3.6
where Kv is a complex flow coefficient.
E q u a t i o n 3.3 also h a s a d e g r e e of a m b i g u i t y in the c o m p r e s s i b i l i t y term

V / B Pa . An a v e r a g e value m a y b e t a k e n for V (volume o f oil u n d e r c o m p r e s s i o n
P a ) if o n l y s m a l l - s c a l e m o t i o n of the m a s s is envisaged. If the m o t i o n is of large-
scale, V should a c c o u n t for the e x p a n d i n g c y l i n d e r volume due to m o t i o n b y using,
V = V(0) + A X m 3.7
The a s p e c t of a m b i g u o u s e q u a t i o n s e l e c t i o n for d y n a m i c a n a l y s i s is c o ~ o n to all

modelling procedures. It w i l l b e d i s c u s s e d m o r e fully in C h a p t e r 7.
3.5 S E L E C T I O N OF P A R A M E T E R V A L U E S
Before a solution (simulation) can b e made, the v a r i o u s c o e f f i c i e n t s a p p e a r i n g

in the m o d e l m u s t h a v e numbers a s s i g n e d to them. For a p a r t i c u l a r system, some of
the n u m b e r s are r e a d i l y assigned: for the e x a m p l e of Fig. 3.1, L1 , L2 , A , V ,
and Ck should be in this category. However, a s s i g n m e n t o f some numbers can b e
a m b i g u o u s or difficult. In the p r e s e n t case, v a l v e flow c o e f f i c i e n t s Kvl and
Kv2 , or Kv , and friction c o e f f i c i e n t Rf are l i k e l y to be d i f f i c u l t to p i n down,
whilst bulk modulus B m a y be ambiguous - - anywhere in the range 500 to 1750 MPa .
This a s p e c t of d y n a m i c analysis also is c o m m o n to all m o d e l l i n g p r o c e d u r e s and

w i l l be d i s c u s s e d more fully in C h a p t e r 7.
25
3.6 ANOTHER EXAMPLE
Fig. 3.2 a illustrates those components of a hydrostatic drive system which a

particular analyst thinks are significant to the dynamic response of the system.
He wishes to predict the load's speed change ~j(t) due to command changes ~i(t)
to the pump swashplate angle. In forming Fig. 3.2 a , the analyst has
decided to regard the input drive speed to the pump (~p) , as constant
(i.e. as a source). This eliminates primary drive (electric motor and
shaft to pump) dynamics, and the inertia of the pump rotating assembly,
from the analysis;
decided to lump the internal leakages (case drains) of pump and motor and
account for them via the leakage coefficient KZ ;
decided that the resistance of the short hydraulic lines joining pump and
motor do not significantly affect the response ~j(t) ; also that the
return line pressure is so low relative to the drive pressure Pa that it
can be neglected for the dynamic analysis;
neglected torsional compliance of the shafting between motor and load;
lumped the hydraulic motor and its shaft inertias in with the load inertia
~o give the total Ij ;
combined motor and load friction, to be described by the viscous friction

coefficient Rf .
Consider the system to be in equilibrium. When the swashplate angle is changed

by #i radians, the volumetric displacement rate of the pump, Q]~ m3/s , is
described by,
Qp = Kq up #i 3.8
where Kq is the volumetric displacement coefficient for the pump m3/rad/rad ;

i.e. its swept volume per radian/sec angular velocity, per radian swashplate angle
(m3/s per rad/s per tad = m3/rad/rad = m 3) .
The volume flowrate of oil effectively used at the hydraulic motor, Qa m3/s ,
is given by,
Qa = Qp- Q~ - ~ c
=
Qp - K " Pa - V/B Pa 3.9
where Q = K~ "Pa is the internal leakage flowrate,

26
a. Schematic for Dynamic Analysis
[--High Pressure Line Driven Load

_. _ %Pressure Pa, Pa ~j, rad/s-- Ij, Kg'm2--7
~l raa \ Vol~me V, m 3 /
Constant Speed I/ \Leakage Q, m3/s
Drive, rap, rad/s
3 I/ I ~ _[ L
/t , Lino / "
~D'"Pump Make~-up ZHyd , Motor /__
v~ m 3 Ka, rad/m 3
*~' K, N.m/Pa Drive Friction
Rf, N.m
b. Block Diagram for System
Qa
-i ..... I.......................... J ~ ..........
FIG. 3.2 HYDROSTATIC TRANSMISSION

27
Qc = V / B - P a is the flowrate absorbed in c o m p r e s s i o n o f the oil,

KZ is l e a k a g e flow coefficient, (m3/s)/Pa ,
V is volume o f oil u n d e r c o m p r e s s i o n , m 3 ,
B is b u l k m o d u l u s o f oil, Pa .
Motor s p e e d mj due to Qa is g i v e n b y
ej = Ka Qa 3 .i0
where Ka is the g e c ~ e t r i c flow s p e e d c o e f f i c i e n t for the motor, rad/m 3 ;

i.e. rad/s p e r m 3 / s .
The m o t o r p r e s s u r e - t o r q u e r e l a t i o n s h i p is
La = K Pa 3 .ii
where K is the m o t o r ' s p r e s s u r e - t o r q u e coefficient, N m/Pa .
A p p l y i n g N e w t o n ' s l a w to the d r i v e n i n e r t i a ,
L a - Rf ~j = Ij ~9 3.12
where Rf ~j d e s c r i b e s the v i s c o u s frictional torque w h i c h r e s i s t s m o t i o n .
The set o f e q u a t i o n s 3.8 to 3.12 is the m o d e l p r o p o s e d for the s y s t e m in o r d e r

to study the d y n a m i c r e s p o n s e ~j(t) of the m o t o r and load to s w a s h p l a t e a n g l e
changes #i(t) . The e q u a t i o n s are in an i n f o r m a l l y s t r u c t u r e d form. T h e y are
readily c o m b i n e d to p r o v i d e a t r a n s f e r f u n c t i o n o f the f o r m
~_i
~i = K2 " D z + K3 D + i 3.13
where the K's are c o m p o s e d o f system coefficients, and D is o p e r a t o r d/dt .

Alternatively, a b l o c k d i a g r a m is r e a d i l y constructed, Fig. 3.2 b .
The e q u a t i o n s are too simple to w a r r a n t e x p r e s s i o n as a vector-matrix model

unless the a n a l y s t p r e f e r s to solve the m o d e l from its v e c t o r - m a t r i x form.
If it is d e s i r e d to study the system in a p o s i t i o n control m o d e r a t h e r than the

speed control mode, w e simply add the e q u a t i o n
8j = I~j a t + @j (0) 3.14
where 8j is a n g u l a r p o s i t i o n of the load. This leads to the t r a n s f e r function,

28
3.15
and a corresponding change in the block diagram of Fig. 3.2 b . This formulation
makes sense if swashplate angle ~i is controlled from neutral, through a desired
input cycle, and back to neutral, causing the load to move from one equilibrium
position to another.
Note: The physically common notion of angular velocity in revolutions/minute or

revolutions/second, can be used for pump and response speeds, ~p and ~j , if
velocity coefficients Kq , Ka , Rf are suitably adapted. However to avoid
errors in designation of numbers in models, the base SI unit radian/second should
be used in all calculations and coefficients.
3.7 CONCLUSION
Conventional modelling procedures used in hydraulic control systems analyses

have been reviewed briefly on the basis that the reader needs a reminder or
orientation only. Some references which explain or apply these techniques more
fully to hydraulic control systems follow.
3.8 REFERENCES
3.1 Blackburn, Reetof, and Shearer, Fluid Power Control, Tech. Press M.I.T.,
1960.
3.2 Lewis and Stern, Design of Hydraulic Control Systems, McGraw-Hill, 1962.
3.3 Khaimovich, E., Hydraulic Control of Machine Tools, Pergamon, 1965.
3.4 Walters, R., Hydraulic and Electrohydraulic Servo Systems, Iliffe, 1967.
3.5 Merritt, H., Hydraulic Control Systems, Wiley, 1967.
3.6 Aizerman, M., Pneumatic and Hydraulic Control Systems, Pergamon, 1968.
3.7 Guillon, M., Hydraulic Servo Systems, Butterworth, 1969.
3.8 Kellar, G.R., Hydraulic System Analysis, Indust. Pub. Co., 1970.
3.9 McCloy, D. and Martin, H., Control of Fluid Power, Longman, 1973.
3.10 Prokes, J., Hydraulic Mechanisms in Automation, Elsevier, 1977.

CHAPTER 4
POWER FLOW MODELLING
4.1 INTRODUCTION
The intention here is:
to discuss the case for modelling hydraulic control systems via study of the
instant (dynamic) power flows taking place between the components of the
systems;
to introduce and demonstrate the power port description of hydraulic

components and systems;
to discuss the mathematical description of dynamic power flows between

components, and in the system;
to demonstrate the formation of a power port model for a particular hydraulic

system.
Those readers familiar with these concepts may prefer to pass over or scan this
Chapter and move directly into power bond graph formulation. Those for whom power
flow modelling is relatively unknown should study this Chapter as an introduction to
the more sophisticated power bend graph material.
Generally, hydraulic control systems are contemplated only where substantial

amounts of energy have to be utilized to do work; that is, where large forces or
torques have to be developed to move resisting loads. The amount of work done on
the resisting load is dictated by the amount of energy utilized effectively. The
rate at which the driven load moves is dictated by t ~ Mate at which it absorbs
SO
Power
i Electric PowerTransformers Power
Absorbing
Supply Load
Eieetric Hydraulic
Pum
Hydraulic
C~
I
Inertia
Load
IElec.Power,~EM,Mech.P o w ~ H v d . PowerL, HC ,Mech-
l(Hyd'iLine~
I
PowerS"1
[ I IPwerLines)~I(Shaft)-I
PowerTransmitters
FIG. 4.1 POWERCHAINFOR HYDRAULICCONTROLSYSTEM

3~
energy. A low energy-utilization rate means a slow motion imparted to the load; a
higher energy-utilization rate causes faster motion. Thus, the dynamic response of
a hydraulic control system is dictated by the dynamic energy flows (i.e. energy flow-
rates) taking place.
In engineering, flowrate of energy is called power. That is, so many joules/

second (watts) describes power. Power is a word with many facets, as a glance at a
dictionary will show. One widely accepted and descriptive concept is that power
flows. An electric motor draws energy at a required rate from the electric power
supply mains. That is power flows from the mains to the motor and subsequently to
the load being driven by the motor. A hydraulic pump can draw power from its prime
mover (electric motor, diesel engine, aircraft engine, etc.) and transform it to
hydraulic power capable of being utilized. Fig. 4.1 illustrates a simple hydraulic
control system in which the electric motor, the hydraulic pump, and the hydraulic
power cylinder, act as intermediaries between the source of power (the electric
mains) and the load which requires moving. When the system is activated, power
~0~8 from the mains through the components and to the load. In this s e n s e the term
power flow is used to describe energy flowrate. Power flow will be static if its
associated energy flowrate is constant. Power flow will be dynamic if its energy
flowrate is time dependent.
Accepting the idea of power flow, it follows that system dynamic response is
dictated by dynamic power flow. From this arises the concept that dynamic response
should be predicted through study of dynamic power flow in the system. This idea is
reinforced by realization that a hydraulic system ~ o e 8 function via dynamic power
interchange between its hardware components. The need then becomes to be able to
describe the instant power flows taking place between the assembly of components
which together comprise the system. The components of the system of Fig. 4.1 can be
considered to be a power source (the mains supply); power transformers (the electric
motor which accepts electric power and puts out mechanical power, the hydraulic pump
which transforms mechanical input power to hydraulic power, the power cylinder which
accepts hydraulic power and puts out mechanical power to the load); power transmitters
(the shafting connecting pump to electric motor, the hydraulic line connecting cylinder
to pump); and a power absorber (the driven load).
Further, hydraulic control systems are controlled by very low powered manual or
automatic command signals, and can consequently be regarded as power amplifying in
addition to power transforming and power transmitting systems.
Let us accept the natural prima facie case for investigating the dynamic response
of hydraulic control systems through study of the power flows which take place
following a command input to the system. The benefits of doing this will emerge in
32
due course. Meanwhile, it is necessary to develop the idea further by considering

the nature of the power flows. The basic requirement of dynamic analysis remains
that a mathematical model of a proposed system be formed and then simulated (solved)
to provide the required response information. Thus, we propose to seek a dynamic
mathematical model through consideration of dynamic power flows.
4.2 POWER PORTS
Each component of a hydraulic control system has only a limited number of ways
through which power can be taken in or given out. Neglecting all losses for the
moment, an electric motor can take in power from the mains supply, and can give it
out to its output shaft. That is, it has only two physical power flow connection
points. These are known as power ports, or simply as ports. Fig. 4.2a illustrates
the motor as a schematic, and as a block having two power ports (column 3).
A hydraulic accumulator has only one port through which it can accept or give
out power. It is represented in Fig. 4.2b as a schematic, and as a block having a
single power port.
A hydraulic pump can also be regarded as essentially a 2-port device, the ports
being the drive shaft and the discharge port. However, the internal leakage (case
drain) of a pump affects its power output and consequently the dynamic response of
the system. It should be accounted for in the power port structure as a third port,
through which power is lost, as illustrated on Fig. 4.2c, column 3.
The 4-way control valve is a 4-port device, with power able to flow into and out
from it via the supply port, the two control ports, and the exhaust port. Fig. 4.2d
illustrates this.
A hydraulic llne is an obvious 2-port component, with power in or out only at

its two ends (Fig. 4.2e). A shaft is similarly a 2-port device (Fig. 4.2f).
A double-acting hydraulic cylinder is functionally a 3-port device, as

illustrated on Fig. 4.2g. It receives hydraulic power at one port, returns some of
it to tank through a second hydraulic port, and gives out power through its piston
rod.
A well designed and well placed system reservoir (tank) rarely affects a
system's dynamic response. However, it receives power from the pump leakage, and
from the return line from the control valve. A reservoir can be shown as a 2-port
device receiving these two power flows (Fig. 4.2h).
Power Port Schematic
Co.,~onent
Ideal With Losses
a. Induction
elec. mech.
Electric Motor
i:~wer--~power
in out
T, friction, windage
b. Hydraulic
~ thermal
Accumulator
loss
hydraulicpower,
in and out
c. Hydraulic Pump pump friction,
mech. ~
power - - ~
in ~
hyd.
power
out
~ nert/a
losses
case drain
leakage
d. Control Valve control
supply-~ port 1
tank control
port 2
e. Hydraulic Line hyd. __ hyd.

power--~-i~wer
in out
f. Shaft me ch. me ch.

power ~ power
in out
g. Hydraulic
Cylinder supply-~--~loa d
tank--4.-~
friction,
entry losses
h. Reservoir return flows
I I
i. Inertia Load friction loss
me~.
wer--D
in
FIG. 4.2 SOME POWER PORT CONFIGURATIONS

34
A driven load of the kind illustrated on Fig. 4.1 is ideally a 1-port component,
this being the mechanical connection whereby the actuator is connected to the load
(Fig. 4.2i).
Thus, each component of a hydraulic control system can be regarded as an n-port

device, with n usually between 1 and 4. Significantly, each designated port is an
obvious physical reality of the component.
Basic power port designations are readily expanded to include loss and other
effects which are thought, in a particular case, to affect a system's dynamic
response. For example, an electric induction motor loses power to speed slip and
should be regarded as a 3-port component. It may also lose significant power to
internal friction and windage and could then be regarded as a 4-port component
(Fig. 4.2a, column 4). Similarly, the n-port designation of a hydraulic pump
could include an environmental port allowing for friction loss (Fig. 4.2c, column 4).
Fig. 4.2, column 4 includes some power loss allowances for other devices. The
designer-analyst should not feel compelled to use the most complete power pert
representation. Rather, he should include only those power flows which he feels
will significantly affect the dynamic response of his particular system. The
simplest model adequate for the purpose in hand should always be sought.
Power port blocks can be used to form a power flow schematic of a system.
Simply couple the blocks as adjacent components are themselves coupled. Fig. 4.3
shows a power flow schematic for the system of Fig. 4.1, expanded to include a
control valve and the reservoir, both of which manipulate or receive power in this
system. A power port schematic provides a structure from which the set of equations
which is the model of the system can be prepared.
In surmnary a power port schematic for a hydraulic control system is readily

formed by:
considering the power flowing into and from each component to be used in the
system, and forming a set of power port blocks for the components. These
component blocks are quite reuseable;
simple joining of the component blocks to form the system power flow
schematic. Such joining is directly analogous to physical connection of
the hardware components to form the system. For example, the power output
end of the electric motor power port is made one with the power input end of
the hydraulic pump block. The line joining the two blocks is precisely
analogous to the shaft which connects the two physical components. In both
cases the connection is a power bond.
35
a. Detailed S c h e m a t i c
constant
Voltage
Source
b. Shaft C o m p l i a n c e R e c o g n i z e d
EM -F---
..... Iv
C. Pump Speed A s s u m e d C o n s t a n t
Constant
Speed Line ~ m m
Source
~I Reservoir
d. C o n s t a n t P r e s s u r e Supply to C o n t r o l V a l v e
|
Line
Control
HC Im
Sourj Valve
~ Hyd.
- I sine I~ I
llj
FIG. 4.3 POWER PORT SCHEMATICS
38
The power port schematic is readily expanded to allow for effects thought to
affect system response. For example, if it is required to include the compliance of
the shaft between electric motor and pump, place a block labelled SHAFT in the line
joining the motor and pump blocks (Fig. 4.3b).
Also, a proposed schematic can be simplified to neglect an aspect considered to

be insignificant for a particular response analysis. For example, the dynamics of
the electric motor may have a negligible effect on the response being studied. In
this case the motor is regarded as a source of constant speed, and its block is
removed from the schematic (Fig. 4.3c). In essence, the source of constant voltage
(mains supply) is replaced by a source of constant speed (but not constant torque~).
Fig. 4.3d shows a further simplification, justified in some analyses, in which the
pump dynamics also are neglected, making the pumping system a source of constant
pressure. In each of these source simplifications it is not implied that power is a
source, (i.e. independent); rather, that a particular variable is a source (i.e.

independent). The associated power variable can vary. Finally, one or more of the
system's hydraulic lines can be removed from consideration in the model simply by
removing its block from the power port schematic.
Development of the model from a power port schematic will be demonstrated in

Section 4.4.
4.3 DESCRIBING POWER FLOW
Power flow magnitude is described by the product of two simultaneous variables.

For example,
Electrical power flow is Voltage drop Current
Mechanical power flow is Net force (or torque) Velocity (or angular
velocity)
Hydraulic power flow is Pressure drop Volumetric flowrate
Neat transfer power flow is Temperature drop Heat conductance
Note that
the first variable of each product is a potential variable; i.e. it offers

the potential to induce power flow,
the second variable of each product is a flowrate variable (current is

coulomb/second or ampere, velocity is metre/second or radian/second, etc.).
Fig. 4.4a illustrates a simple hydraulic control system; Fig. 4.4b includes the
37
power flow product variables, with the potential variables above and the flowrate
variables below the component symbols. Arrows indicating the natural flow direction
of power are included - - p o w e r is drawn from the electric mains supply (source of
voltage) and is delivered to the driven load- Neglecting all losses, power is
conserved and
Voltage - Current = Force Velocity
The relationship between the potential and the flowrate variables of a particular
power product is of a cause-effect form. For example, force is the cause w h i c h
induces velocity of the load mass. Voltage applied to an electric motor is a cause,
and the resulting current drawn from the supply is an effect induced by the load
impedance seen by the motor.
At first sight it might appear that the potential variable is always cause and
the flowrate variable is always effect. This is not the case. It is not always
obvious which variable of a power product is cause, and which is effect. Appreciation
of cause-effect is a great aid to the understanding of a system's dynamic behaviour.
It will be demonstrated in later Chapters that assignment of causality (i.e. decision
on which variable of a power product is cause and which is effect) can be a key link
in obtaining from a power flow diagram the correct set of dynamic equations for
simulation.
Consider the system of Fig. 4.4a and b. It is proposed to decide which variable
in each power product is cause and which is effect. We shall rearrange Fig. 4.4b to
show cause variables above and effect variables below, rather than the Fig. 4.4b
arrangement of potential variables above and flowrate variables below. Further, we
shall designate the cause variable by an arrow in the direction of desired power flow,
and the effect variable by an arrow directed against the power flow. This is a
convention. However, it has physical significance, as the effect variables are back
effect8 dictated by the dynamic nature of the system. For example, applied voltage
is a source or constant and can only be designated as a cause as its magnitude is not
affected by the load effects of the system. Designation of voltage as cause
automatically requires current to be designated as an effect. However, this is also
physically the case. The current drawn by the electric motor depends on the load
applied to the motor by t~9 8yst~n, i.eo the current i8 a back effect:
AS discussed earlier, force applied to the load is readily perceived as a cause,

and the subsequent velocity of the load is a consequent effect. There remains some
ambiguity on this interpretation which bond graphs in later Chapters will clarify.
This leaves cause-effect decisions to be made for the hydraulic pump and for the
38
a. Power Flow
Elec. P o w e r ~ M~ch% Power ~yd. Power ~ech.

(cyl. P ~ [ ~
(Wiring)- I Motor I (Shaft) ~I Pump (Hyd. Line) ~ ....
b. Potential and Flowrate Variables
Potential Variables
/ \
Voltage Torque Pressure Force
HC ~ Im
I Shaft - Hyd. Line-- Cyl. Rod
Current Angular S p e e d Flowrate Velocity
\ ...................
Flowrate Variables
.c. Cause and E f f e c t V a r i a b l e s
Cause Variables ......

/
Voltage Speed Flowrate Force

..... T
Power I HP HC ...... ~ Im
Blow ~ EM
...i
.
Current Torque Pressure Velocity
\ /
Effect Variables
FIG. 4.4 POWER FLOW AND ITS V A R I A B L E S

39
hydraulic cylinder. It is not necessary to consider components in the order in which

they appear in the system - - doing this could make the task harder. Let us simply
think about the action of the components, and decide the more obvious causalities.
The remaining causalities will flow from these decisions. The system of Fig. 4.4 is
not a constant pressure system. The pressure developed by the pump is load dependent.
Thus, pressure is a back effect, and can be so designated on Fig. 4.4c, making flowrate
a cause. The torque required at the pump shaft is also load dependent (pressure
dependent if you like). Consequently, torque is also a back effect - - an effect
variable - - m a k i n g motor pemp speed a cause variable. In this manner we complete
the causality assignment for the system.
Check to see that physical sense is not violated. Voltage is applied to the
electric motor, whose motion is resisted by the system load; this requires current
to be drawn from the supply main. Resultant speed of the motor is transmitted
directly to the hydraulic pump, which, being of positive displacement, causes positive
flowrate of liquid to the cylinder. Liquid pumped positively into the hydraulic
cylinder, whose piston motion is resisted by the inertia load, creates rapidly
increasing pressure. This pressure acts on the piston to generate a force, which
acts on the inertia load. The inertia load is put into motion. The magnitude of
the inertia load dictates its acceleration and consequently its velocity, which is
also the velocity of the piston. Piston velocity affects the rate of pressure
change. Pressure dictates the torque required to drive the pump and thus the torque
output required of the electric motor. The torque demanded of the motor dictates
the current drawn by it. These actions do not take place in complete time sequence
as implied here. They are occurring simultaneously, but are out of phase in the
time domain. That is, it takes time for an action to fully develop following an
applied cause. Predicting the sum total of these time effects is what dynamic
analysis is all about.
Note that the effect variable does not necessarily result directly from its own
cause variable. In the present example, this correlation occurs only for the inertia
load, for which velocity is the effect of the causing force. Certainly current is
not caused directly by the applied voltage; it is dictated by the shaft torque
demand.
It cannot be assumed that the causality of a particular power variable will be

the same in all systems and analyses. For example, it is sometimes appropriate to
assume pressure a cause, and flowrate an effect. In constant pressure supply systems
this will be the case. Making causal sense in a particular system is a great aid
towards understanding the dynamic nature of the system, and in development of its
dynamic model.
40
4.4 GETTING EQUATIONS
So far, we have reached the stage where a power flow schematic can be proposed.
Thinking this out and making the associated decisions provides insight into those
aspects of its components which will affect a system's dynamic performance. The
ultimate objective remains the production of a set of e Uuations (the dynamic model)
which can be ~imulated (solved) to obtain the predicted behaviour of the significant
system variables. It will become apparent that a power flow schematic can provide
the structure from which a complete and suitably arranged set of equations can be
written.
It is necessary to write expressions relating the variables and coefficients

associated with e a c h p o w e r bond (the line joining connected power ports). The
relationships available are those utilized in all dynamic modelling procedures - -
i.e. the basic physical laws and relationships describing the various actions which
take place when a system is activated.
For example, consider the simple situation of Fig. 4.5a, for which supply
pressure Ps is regarded as constant (a source). When the variable position 4-way
closed-centre control valve is actuated by the displacement input signal Xv , the
system will respond dynamically to move the inertia load. Fig. 4 . 5 b s h o w s the
initial power port structure proposed to describe dynamic power flow following
application of Xv . Hydraulic line resistance has been neglected. If power loss
in the return line from cylinder through valve to tank is considered to be negligible
as far as load dynamic response is concerned, the power flow schematic can be
simplified to that of Fig. 4o5C. Each power bond (line) on Fig. 4.5c is allocated
its power product variables, as follows:
Ps Qs describes power flowing from source to control valve,
Pa ' Qa describes power flowing from control valve to cylinder,
~.~m describes power absorbed by the load (i.e. flowing from cylinder to
load.
where Ps is the supply pressure, Qs the supply flowrate;
Pa is the pressure developed in the cylinder, Qa the effective flowrate to

the cylinder - - i.e. the volumetric displacement rate of the piston in
the cylinder;
Fm is the net force applied to the driven load, and ~ the velocity
developed by the load.
Also, we propose to account for a power loss due to friction associated with the
41
a. The System
xv1 Hyd. Cylinder,

Load,
Area A
Constant
Pressure I t
Supply (Ps) | iI
Tank Xm
Valve
b. Initial Power Port structure Proposed
Friction
Source
Ps ~ Cont. i
Valve |~
iTM
c. Simplified Power Port Structure, with Power Variables on

Each Power Bond
Sourcel P~
Ps I Qs
FIG. 4.5 POWER PORT STRUCTURE FOR SIMPLE HYDRAULIC SYSTEM

42
driven load, and which is described by Ff " ~ , where Ff is the force required to
overcome friction.
Being satisfied that the power flow schematic of Fig. 4.5c is adequate for our
purpose, we now require to prepare from it an appropriate and complete set of
equations. This requires an equation for each power variable, expressed in terms of
only other system power variables and system coefficients. Further equations which
might be required to describe variables other than these po~Jez~ 8tatg UGI~GbZe8 can be
added if desired. Environmental power loss effects, such as load friction in the
present case, can be described as separate equations, or can be built into the power
state equations, whichever the analyst may prefer. For the present case, we can
write the six power state equations:
Ps = constant (an assumption) 4.1
Qs = f(xv, (Ps - Pa)) (orifice flowrate, form to be decided) 4.2
Pa = - B/V I(Qs - Oa)dt + Pa(O) (capacitive effect) 4.3
Qa = A ~ (piston swept volume rate) 4.4
Fm = Pa A - Ff (net force on Im ) 4.5
= i/Im (Fm dt + ~ ( 0 ) (Newton's law, integral form) 4.6

J
We look now for any additional equations required
to specify any significant response variables other than the power state
variables,
to specify any non-constant coefficients used in the state equations,
to specify the controlling input.
In the present example, this leads to the auxiliary equation set:
xm -- I ~ at + Xm(O) 4.7
( )~ is a respone variable of primary interest)
V = V(O) + A Xm 4.8
( V is volume of oil under compression)
Ff = to be decided 4.9
(friction force)
43
Xv = to b e s p e c i f i e d 4 .I0
(control valve d i s p l a c e m e n t input)
Coefficients in the e q u a t i o n s not a l r e a d y identified are:
A effective area of hydraulic ram
B - bulk modulus o f the fluid i n the s y s t e m
V - v o l u m e of fluid u n d e r compression Pa (i.e. v o l u m e in hydraulic line

from valve to c y l i n d e r p l u s the c y l i n d e r chamber). V will v a r y w i t h
piston displacement and can be e x p r e s s e d V = V(O) + A Xm .
These e q u a t i o n s will be readily recognized b y those familiar with dynamic

analysis o f h y d r a u l i c systems. The precise relationship o f e q u a t i o n 4 . 2 can b e
chosen from several w i d e l y used forms, including:
Qs = Kvl Xv - Kv2 Pa 4.2a
a linear e q u a t i o n w i t h Kvl and Kv2 as flow a n d p r e s s u r e coefficients respectively;

orl
Qs = Kv Xv sgn (Ps - Pa) - (Ps - Pa) 4.2b
a non-linear flow d e s c r i p t i o n with Kv an orifice coefficient.
Similarly, the a n a l y s t can c h o o s e the p r e c i s e friction force e x p r e s s i o n from

forms such as
Ff = Kf Xm 4.9a
which d e s c r i b e s linear v i s c o u s friction of coefficient Kf ; or,
Ff = s g n Xm - Fu + Kf Xm 4.9b
which i n c l u d e s a coulomb friction term Fu .
It is quite an a d v a n t a g e to be able to d e f e r d e c i s i o n s on the forms of some o f

the m o r e ambiguous relationships to be u s e d in the final simulation.
The s e t o f e q u a t i o n s 4.1 to 4 . 1 0 is the d y n a m i c m o d e l p r o p o s e d for the s y s t e m o f
Fig. 4.5. The m o d e l is c o m p r i s e d o f a set o f quite simple b a s i c r e l a t i o n s h i p s in

algebraic and integral form. Note that Xm i s a state v a r i a b l e , a n d t h u s its
appearance in an e q u a t i o n does n o t m a k e that e q u a t i o n o f d i f f e r e n t i a l form - - it is
not i n t e n d e d that Xm be d e r i v e d from Xm .
44
Actually, the selection and arrangement of equations is not as clear cut as

appears here. There are some causal decisions to be made in order to produce
integral rather than differential expressions. For computer simulation it is far
preferable to have equations in integral rather than derivative forms. In the
present case, equations 4.3 and 4.6 reflect such causal decisions, and these affect
the arrangement of the other equations. A power flow variable should be the object
of one equation only, and each power flow variable is to be the object of an equation.
A more substantial system or analysis would result in a much larger number of

equations being required, and considerable confusion on the equation arrangement, and
even on completeness or otherwise of the equation set, can arise. Unruh [4.1] and
Young [4.2] have separately developed procedures for formally sorting and arranging
equations from power port structures to satisfy both causality and completeness.
Selecting of equation forms, and the obtaining of numerical values for coefficients
are discussed more fully in Chapter 7.
4.5 AN EXAMPLE
The following example is adapted from Reference 4.3.
Fig. 4.6a illustrates a hydrostatic drive system whose dynamic response and
performance characteristics are required prior to purchasing the components and
constructing the system. Fig. 4.6b shows a power port diagram prepared for the
system. Each line between blocks is a power bond, representing a physical power-
transmitting connection. Each line has two power state variables associated with
it, their product describing the power flowing. At first sight it appears that
there will be 2n power state variables, where n is the number of power bonds
present. However some bonds share a power state variable. For example, Pp is
the pressure on the bond between pump and relief valve, and also on the bond between
relief valve and filter. Also, ~a is the velocity (flowrate variable) on both the
hydraulic motor-load bond and the load friction power dissipation bond. Examination
of the power port diagram reveals the following set of power state variables, starting
from the assumptions that the electric motor runs at constant speed, and that the
charge pump power requirement does not affgct the dynamics of the main system:
S~ constant speed of electric motor, and hydraulic pump

T torque developed in the motor-pump shaft
Pp pressure developed by the pump
Qp flowrate developed by the pump
Pe reservoir pressure, often assumed zero
Qs flowrate from reservoir to pump
QEp leakage flowrate from pump
Qr flowrate through pressure relief valve to reservoir
Qb flowrate into filter (also from filter)
Pb pressure at discharge side of filter
4S
a. Circuit Diagram
~'tric Motor
Pressure Compensated Pump
Low Pressure Relief Valve

F" High Pressure Relief Valve
EM
From Tank
"" ~-- ~ u--Crossline Relief Block
~arge Pump i._ D i r e c t i o n a l Control Valve

Relief Valve
b. Power Port
IElectric I
Motor I
T S~ Tfa ~a
Pe~QZp PelQr ~ Valve ~ Blck ~ ' = ~ !
FIG. 4.6 A HYDROSTATIC DRIVE AND ITS POWER PORT DIAGRAM

46
Pal pressure developed in line between control valve and hydraulic motor
Qvl flowrate from control valve towards motor
Qal flowrate received by hydraulic motor
Pa2 pressure developed in return line between motor and control valve
Qa2 flowrate from motor towards control valve
Qv2 return flowrate entering control valve
Qe discharge flowrate from control valve to reservoir
Qre flowrate from crossline relief block to reservoir
Prc pressure in charge line to crossline relief block
Qrc flowrate in charge line to crossline relief block
Qia leakage flowrate from hydraulic motor
Ta torque developed in motor-load shaft
~a angular velocity of motor and load
Tfa friction torque
To form the model, an equation for each of these 24 power state variables is
required. If numbers can be provided for the various coefficients, the set of
equations can be simulated to provide the sought-after dynamic response information.
The system can be controlled or disturbed by adjusting the state of the variable
delivery pump, by adjusting the 4-way control valve, or by adjusting load inertia or
friction. This example and its equations are developed more fully via bond graphs
in Chapter 8.
4.6 SUMMARY
The essential components of a hydraulic control system can be placed in one of

three groupings:
power transforming components (electric motors, pump, actuators, etc.),
power transmitting components (electric supply wiring, shafts, hydraulic lines,

cylinder rods, etc.),
power control components (pressure and flow control valves, etc.).
The dynamic response of a hydraulic control system is dictated by the dynamic

power interchanges taking place between the components of the system.
Power can enter or leave a component through only a small number of readily
identified and physically meaningful power ports. Adjacent components are
connected together functionally only via these power ports.
A dynamic model of a system, expressed as a set of simultaneous action equations,

can be developed from power flow considerations. A power port block diagram is
first formed to provide the structure from which the equation set will be
prepared. Each line (power bond) joining power port blocks is used for power
interchange.
47
Power flow requires description as the product of two variables.
Each power variable is a power state variable which is the object of one of the
set of power state equations which comprise the basic dynamic model of the system.
Using preferred causality, the power state equations can be arranged in algebraic
and integral forms suitable for computer solution or simulation.
System power port structures are modular and closely resemble the modular
construction of the hydraulic system itself. This allows components to be
readily changed, added, or deleted from a system's power port block diagram.
Dynamic coupling effects are inherently included in mQdels derived from power
port structures.
The advantages relative to conventional modelling procedures of the foregoing

points are quite significant.
The development of power port models will not be pursued further, as the Power
Bond Graph techniques to be developed in the next Chapters are a more advanced and
formal development of power flow modelling.
4.7 REFERENCES
4.1 Unruh, D., A Standard Format for Mathematical Models of Fluid Power
Systems, Proc. N.C.F.P., V26, 1972.
4.2 Young, M.Y., Digital Simulation of Hydraulic Control Systems, M.Eng.Sci.

Thesis, Monash University, Australia, 1973.
4.3 Stecki, J.S. and Cuthbertson, S.C., The Effects of Changeover to Fire
Resistant Fluilds on Hydraulic Performance - Part 2: System Criteria,
FRH Jnl., VI, N2, pp 167-173, FPRC, Oklahoma State University, 1981.
CHAPTER 5
POWER BOND GRAPHS
5.1 INTRODUCTION
So far it has been proposed
that prediction and subsequent optimization of dynamic response should be

part of the procedure for the design of hydraulic control systems,
that inadequate availability of dynamic models in forms suitable for use by

the designer-analyst is the main obstacle,
that conventional model forms are not particularly suitable for use in the
system design situation,
that dynamic power flows are particularly descriptive of the physical actions
dictating the dynamic response of hydraulic control systems,
that power flow models offer the system designer modular reuseable model
forms which are close to direct representation of the similarly modular
hydraulic system,
that power flow modelling allows increased separation of the essential steps,
and consequent increased formality, in forming a set of dynamic equations
(the model) suitable for solution by computer.
The present Chapter introduces the power bond graph technique for developing
dynamic models of powered control systems. It has become customary to refer to
power bond graphs simply as bond graphs. Although the present author considers this
to be an undesirable abbreviation, it is adopted for the Text. Bond graphs have
49
been applied to a wide variety of dynamic s i t ~ t i o n s , including vibration analyses,

thermodynamic and fluid dynamic situations, and physiological systems. However,
they are used in this Text solely as applicable to hydraulic control systems, for
which they have a quite particular affinity. Only those aspects of bond graphs
which have been found to be useful in the hydraulic control systems area will be
discussed and utilized. This includes representation of electric motors, electric
servovalves, inertia loads, and other such devices which occur within the field of
hydraulic control systems. References 5.1 to 5.4 provide a more historical and far
more complete information on bond graphs and their applications.
A bond graph is a dual-signal flow diagram formed from a limited set of symbols.
From it a properly ordered and arranged set of dynamic equations can be prepared in a
form suitable for computer solution. A new but very limited set of terms and
symbols has to be learned. In effect, the multiport power flow approach of
Chapter 4 is made more general, more concise, and more formal through the (power)
bond graph symbols. We shall begin by introducing and discussing the basic terms
and symbols which w i l l b e used in subsequent development of bond graph derived models.
The reasons why bond graph techniques are being proposed in this Text as superior in
the design-orientated analysis area will be presented, discussed, and demonstrated
after the reader has gained an appreciation of what bond graphs are.
The strategy behind use of power bond graphs can be summarized:
to use a very limited number of versatile general terms and symbols to

provide a rational graphical structure which describes the presence, and
the interactions, of effects which affect the dynamic performance of the
system;
to allow ready formation of, and subsequent changes in, the structure as is
required in the creative design situation;
to use the model structure to formally prepare a rational and adequately

complete set of equations suitable for computer simulation of the system.
5.2 BOND GRAPH TERMS AND SYMBOLS
5.2.1 Effort and Flow Variables
In Chapter 4 it was illustrated that power flow is described by the product of

a potential variable and an associated flowrate variable. Power bond graph
researchers and users have accepted the terms 8ffo~t and ~ o w to mean potential and
flowrate respectively. Thus, in a hydraulic line, pressure is the effort variable
and volumetric flowrate is the ~ 0 ~ variable. In this Text, E and Q will be
50
used as the general symbols for effort and flow respectively. In specific situations,
specific symbols will be used; for example p for effort (pressure) and Q for f l ~
(flowrate) in hydraulic line power flow.
5.2.2 Sources
One or more of the power variables associated with a dynamic system can be
constant or can be assumed to be constant for purposes of analysis. For example,
the voltage applied to an electric motor is assumed to be constant even

though the p o w e r drawn from the electricity supply by the m o t o r fluctuates,
some hydraulic control systems are analysed on the basis that supply to the
control valve is at constant pressure,
in most situations, pump drive speed may be regarded as constant.
In bond graph terms, each of these constants is called a 8ou~de, and is designated by
the symbol S - In general, in the present Text,
SE denotes an effort source

SQ denotes a flow source
although specific effort and flow variables will be specifically described. For
example:
SP denotes a source of pressure

SQ denotes a source of volumetric flowrate.
Designation of a power variable as a constant does not imply that the associated
power is constant. In fact, in a dynamic situation power will rarely be constant.
The product variable associated with the designated source can vary, thus allowing
power to vary. For the electric motor, voltage (effort) is constant b u t current
(flow) varies as dictated b y the load on the motor. For a constant supply pressure
hydraulic system, flowrate can vary and power flow will vary accordingly. For power
flow to be constant, either both effort and flow variables must be sources, or both
must be allowed to vary such that their product remains constant. The former case
is not compatible with dynamic states, though the latter may be sought in some
control system designs.
A pc~er state variable which is independent of the system, though not necessarily
constant, is also regarded as a source.
5.2.3 Power Bonds
A path through which power flows is represented by a line known as a p o ~ r bond.

S~
Each power b o n d h a s an e f f o r t and a flow v a r i a b l e a s s o c i a t e d w i t h it. In g e n e r a l it

is represented
E
Q
or, in a p a r t i c u l a r case such as a h y d r a u l i c line,
P
Q
5.2.4 Power Transformers
A s d i s c u s s e d in C h a p t e r 4, m a n y o f the m a j o r c o m p o n e n t s in a h y d r a u l i c control
system are p r i m a r i l y p o w e r transforming devices. Thus, an e l e c t r i c m o t o r c o n v e r t s
electric p o w e r to m e c h a n i c a l power; a h y d r a u l i c p u m p c o n v e r t s m e c h a n i c a l p o w e r to
hydraulic power; a h y d r a u l i c a c t u a t o r c o n v e r t s h y d r a u l i c p o w e r back to m e c h a n i c a l
power. P o w e r t r a n s f o r m a t i o n is r e p r e s e n t e d in b o n d g r a p h s as the symbol TF in the
centre of a p o w e r bond, as follows:
- - TF
and,
E1 E2
Q1 Q2
which m e a n s t h a t
E1 t r a n s f o r m s to E2
Q1 t r a n s f o r m s to Q2
while p o w e r is c o n s e r v e d such that
E1 Q1 = E 2 Q2
For the s i m p l e s t transformations, E1 is r e l a t e d to E2 b y a constant, and Q1

to Q2 b y the i n v e r s e o f the constant. In these c a s e s the t r a n s f o r m i n g m o d u l u s is
noted b e l o w the TF symbol. For example, p o w e r t r a n s f o r m a t i o n i n a h y d r a u l i c
cylinder is d e s c r i b e d by:
P F
TF
Q A
where P and Q are the p r e s s u r e and v o l u m e t r i c flowrate acting in the cylinder;

F and X are the force and v e l o c i t y i n d u c e d on the c y l i n d e r ' s p i s t o n and rod;
A is the e f f e c t i v e c r o s s - s e c t i o n a l a r e a of the piston.
52
The symbol m e a n s simply
P A = F
Note t h a t p o w e r is conserved,
(P - A) - Q/A = P - Q = F - X .
As a n o t h e r e x a m p l e , if a hydraulic p u m p is r e g a r d e d as an ideal p o w e r transformer,

it w o u l d be r e p r e s e n t e d in a b o n d g r a p h as,
T P
Q
vp-i
which means
T V p -I = P
Vp = Q
T w = P - Q
where T (torque) [~ (angular velocity) describes the p o w e r i n t o the p u m p at its

drive shaft,
P (pressure) Q (flowrate) describes the p o w e r d e l i v e r e d b y the p u m p if all
losses are neglected;
Vp is the swept v o l u m e p e r radian o f the pump.
It is e n l i g h t e n i n g to r e a l i z e that in most hydraulic s y s t e m p u m p situations,
speed o f the p u m p d r i v e is i n t e n d e d to b e constant, while torque d e v e l o p e d is

dependent on the p r e s s u r e generated in the system. On this c a u s e - e f f e c t basis, we
might prefer to w r i t e the e q u a t i o n s
Q = t0 - v p
T = P - Vp
i.e. showing the e f f e c t v a r i a b l e s as f u n c t i o n s o f the cause v a r i a b l e s . S u c h causal

decisions w i l l be d i s c u s s e d shortly.
F o r a fixed d i s p l a c e m e n t pump, Vp is constant. However, for a v a r i a b l e
displacement pump, Vp is a f u n c t i o n o f the v a r i a b l e p u m p stroke. This wQuld be
recognized i n b o n d g r a p h t e r m s as the modulated t1~nsfolTner,

53
T P
- - MTF - -
Q
V p -I ()
where () embraces whatever variable(s) control Vp .
Losses and dynamic effects which occur during power transformations are accounted
for separately from the actual TF operation, which is treated as ideal; represent-
ation of losses and dynamic effects in bond graphs will be discussed and demonstrated
shortly.
Gyration
A variation sometimes arises w h e r e i n the input effort variable transforms to a

flow variable, and the input flow variable transforms to the output effort variable.
This power transduction is called ~rUPat~on, and its power bond graph symbol is GY ,
as follows
E1 E2
GY
Q1 K Q2
the implication being that
E1 K = Q2
QI/K = E2
rather than
E1 K = E2
QI/K = Q2
which is the TF case.
The induction electric m o t o r is an example. Applied voltage E (effort)

induces speed (flow) o f the motor field windings; the torque T induced
requires current i to be drawn. In bond graph symbols this is described by
E T
GY
i
K
which means
E " K =
i/K = T
with power conserved such that

54
(E K) i/K = E i = T
Note: is n o t m o t o r shaft speed, w h i c h will b e less than m by the a m o u n t o f

speed slip n e c e s s a r y to d e v e l o p torque in the i n d u c t i o n motor. Further
d e v e l o p m e n t of the b o n d graph o f a n induction m o t o r will be d e m o n s t r a t e d
shortly.
As w i t h TF's , g y r a t o r s can be m o d u l a t e d i n w h i c h case they w o u l d be designated

as MGY w i t h the a p p r o p r i a t e m o d u l a t i n g f u n c t i o n i n d i c a t e d underneath.
5.2.5 Dynamic Effects
We c o m e now to those e f f e c t s w h i c h d i c t a t e the d y n a m i c nature of a control

system and w h i c h g o v e r n its d y n a m i c response. These a c t i o n s can be p l a c e d into one
of the three general c l a s s i f i c a t i o n s ,
Resistive Power D i s s i p a t i o n
C a p a c i t i v e Power Storage
Inertive (or Inductive) Power Storage
The p o w e r b o n d g r a p h symbols for these are r e s p e c t i v e l y
R C I
Ill
That is, a p o w e r b o n d l e a d i n g to one of the three b a s i c d y n a m i c a c t i o n classifications.
E a c h c l a s s i f i c a t i o n will n o w b e d e s c r i b e d through examples.
5.2.6 Resistive Power Dissipation
Any e f f e c t g o v e r n e d b y an algebraic r e l a t i o n s h i p b e t w e e n n e t e f f o r t and f l o w r a ~

can b e c l a s s i f i e d as a ~esiotiu~ effect. R e s i s t i v e effects d i s s i p a t e power.
Resistive e f f e c t s w h i c h can a f f e c t the d y n a m i c r e s p o n s e of h y d r a u l i c control systems
include:
friction in all of its forms. In p a r t i c u l a r , f r i c t i o n a s s o c i a t e d w i t h

d r i v e n loads is a s i g n i f i c a n t factor in the d y n a m i c r e s p o n s e of h y d r a u l i c
systems;
pipe line p r e s s u r e drops. P o w e r flow in a p i p e l i n e is d e s c r i b e d b y the

p r o d u c t o f p r e s s u r e and flowrate. For a pipeline, the f l o w r a t e s a t e n t r y
and exit are e s s e n t i a l l y the same. However, there will be a p r e s s u r e drop,
and c o n s e q u e n t l y a p o w e r loss, b e t w e e n e n t r y and exit;
55
control orifices of various kinds (flow control valves, pressure relief

valves, etc.). These too cause pressure drops at essentially identical
inlet and exit flowrates and so dissipate power.
The examples all illustrate that power dissipation actions are resistive in
nature. Further, they cause irreversible power loss. The bond graph symbol
recognizing a power dissipation action is a power bond with the letter R (for
Eesistive effect), at its end. The R indicates that power transmission in the
region of the b o n d is dictated or affected by a resistive effect. Power flow in the
bond is described by the product o f an effort variable and a flow variable. Thus
the symbol becomes
and its implication, to be developed later, is that a resistive effect R is present

and that it describes a relationship
E = R(QI , or Q = R(P) 5.1
where R() denotes "a resistive function of".
For example, consider the pressure drop in a hydraulic line. It can be

recognized in bond graph terms as
Rh
dPh Qh
If flow is assumed laminar, the linear resistance relationship can be utilized to

provide
Qh = i/Rh - dPh 5.2
where Rh is the laminar resistance coefficient for the hydraulic line.
If, o n the other hand, flow in the line w a s turbulent, the Rh effect cannot be
accounted for via a simple coefficient. The pressure drop-flowrate relationship is
better described b y the non-linear equation
Qh = Kh - sgn APh - Aph % 5.3
where Qh is the volumetric flowrate in the hose;

APh is the magnitude o f the pressure drop across the hose;
58
Kh is a p a r a m e t e r d e s c r i b i n g h o s e geometry, fluid p r o p e r t i e s , and f l o w

coefficients.
I r r e s p e c t i v e of w h a t Rh r e l a t i o n s h i p is e v e n t u a l l y chosen to relate Qh to
APh , the b o n d g r a p h symbol r e m a i n s as shown. The symbol r e c o g n i z e s the action, not
the s p e c i f i c e q u a t i o n to b e used. The symbol i n d i c a t e s that f l o w r a t e is a function
o f p r e s s u r e drop, or vice versa. E q u a t i o n s e l e c t i o n c o m e s later.
F r i c t i o n can be i n c l u d e d in b o n d g r a p h s as a r e s i s t i v e e f f e c t Rf ; for example,

as
Rf
which means
Rf
Ff
where Ff is f r i c t i o n force r e s i s t i n g m o t i o n X .
The i m p l i c a t i o n is that F f = Rf(X) ; that is that friction force is a resistive

f u n c t i o n of velocity. In the s i m p l e s t case o f v i s c o u s (linear) friction, Rf is a
simple c o e f f i c i e n t and the e q u a t i o n i m p l i e d is
Ff
s l o p e Rf Ff = Rf X 5.4
In the m o r e g e n e r a l case in w h i c h c o u l o m b f r i c t i o n (stiction) i s also present,

we m a y n e e d to d e s c r i b e
Ff Ff
in the Rf function. This w i l l b e further d e v e l o p e d in C h a p t e r 7.
A n action o r device p r o v i d e s an R e f f e c t if i t follows the c o n c e p t t h a t effort

d r o p induces flow, o r f l o w is a c c o m p a n i e d b y e f f o r t drop.
57
5.2.7 Capacitive Power Storage
A device provides a capacitive effect if net ~ O W into it causes increased

ef~omt within it. Capacitance implies energy storage and thus power storage
characteristics. Capacitive power storage devices include:
mechanical springs,
drive shafts in torsion,
mechanical compliance in general,
hydraulic accumulators,
electric capacitors,
in general, oil-containing volumes in hydraulic control systems
(filters, hydraulic hoses, cylinder end volumes, etc.).
In general, these devices can store and give up potential energy in manners
affecting dynamic performance of the system of which they are part. The bond graph
symbol for a capacitive effect is:
t
the line (power bond) representing the path through which power can flow to or from
the particular capacitive effect recognized by the C at the end of the bond.
Implied on the bond are the effort and flow variables associated with the particular
C being considered,
and the further implication is that
E = C(Q) or Q = C(E)
where C( ) denotes "a capacitive function of".
In the simplest case of linear capacitance, the function becomes
m = I/C IQ dt + E(O) 5.5
i.e. for an electrical capacitor
E = I/C li dt + E(O) 5.6
(Faraday law, E is voltage, i is current)
i.e. for a linear spring

58
~ = i/Ck IX dt + Fk(O) 5.7
where Fk is the force developed in the spring;

Ck is the compliance of the spring;
is the velocity of one end of the spring relative to the other end.
Note that equation 5.7 is the dynamic form of the static spring relationship
Ff = 1/Ck X
i.e. for linear oil compressibility in a fixed chamber or volume
P = I/C [Q at + P(O) 5.8
where P is the pressure in the chamber;
C is the hydraulic capacitance of the chamber; ideally, C = V/B

for a fixed volume, where V is the volume of oil under compression,
and B is bulk modulus of the oil;
Q is the flowrate of oil into the chamber.
AS w i t h the R effects, there is no need for a designated C to be a

coefficient. In a particular case, the C effect linking a potential (effort) to a
flowrate (flow) m a y be a non-linear expression, or even empirical data. Several C
relationships will be demonstrated later in the Text. Note that w~ have, in each
case, expressed the potential variable as a function of the flowrate variable. This
satisfies the b a s i c causality requirement that the effect is a result of the cause.
The capacitive e f f e c t is referred to as "power storage". This does not mean

that power is always flowing into a C element. In dynamic action, power can be
flowing f~om the C element into the system. That is, power flow can be positive
or negative. This does not affect bond graph development; sign conventions will
introduced later.
To be classifiable as a C effect, an action or device should follow the concept

that the flow into a C effect causes effort build up w i t h i n it.
5.2.8 inertive Power Storage
Inertive power storage o c c u r s w h e n inertias are accelerated, storing u p kinetic

energy. The power bond graph symbol is:
59
with the power bond (the line) indicating the means through which power may be given
to the inertia, or given up by it. Adding the ~ffort and the flow variable
associated with the possibilities
linear motion rotary motion
leads to
Im ij
Fm ~ or TjI~J
(linear motion) (rotary motion)
whe re Fm is the net force applied to mass Im to cause dynamic linear velocity
Xm of the mass;
Tj is the net torque applied to rotating inertia Ij to cause dynamic

angular velocity ~j .
The relationship implied on the I power bond is, of course, Newton's law.
However, it is best expressed as
= I(F) or ~j = I(Tj)
rather than the more usual
Fm = I(X) or Tj = I(~j)
In fact, when we realize that force (torque) is cause, and velocity (angular velocity)
is effect, it is more appropriate to express Newton's law as:
Xm = I(Fm) or mj = I(Tj)
and this leads to
= i/Im IFm dt + ~ ( O ) 5.9
or
f
~j = i/Ij )Tj dt + ~j(O) 5 .i0
as the dynamic relationships implied on the respective Im and Ij power bonds.
Inertia effects are referred to as "power storage". As with capacitive effects,

I effect power flow can be positive or negative, with the inertia absorbing or
giving up power respectively. This will be discussed in Section 5.4.
60
E l e c t r i c a l i n d u c t a n c e is a n o t h e r d y n a m i c I effect. It is d e s c r i b e d b y a
d y n a m i c e q u a t i o n similar in form and nature to N e w t o n ' s law, i.e.
E = L di/dt , or i = 1/L IE dt
w h e r e voltage E and c u r r e n t i are c l e a r l y the e f f o r t and flow v a r i a b l e s

respectively, and are r e s p e c t i v e l y a n a l o g o u s to force and v e l o c i t y in N e w t o n ' s law.
The general c o n c e p t o f an I e f f e c t is that flow is d e p e n d e n t on the time

i n t e g r a l o f effort.
5.2.9 Summing Junctions
A p o w e r b o n d graph, like all signal f l o w diagrams, r e q u i r e s summing junctions at

w h i c h v a r i a b l e s can b e a d d e d algebraically. A p o w e r b o n d g r a p h is, however, a dual
signal d i a g r a m - - each p o w e r b o n d in it "contains" the two s i m u l t a n e o u s variables
w h o s e p r o d u c t d e s c r i b e s the p o w e r flow in the bond. A s o n l y like v a r i a b l e s can be
added to or s u b t r a c t e d from e a c h other, i t is n e c e s s a r y to have two d i s t i n c t k i n d s of
summing junctions in p o w e r b o n d graphs. One o f these, d e p i c t e d b y O and called an
O-JUNCTION, r e c o g n i z e s summation o f the flow variables, as follows:
E Q2
I
E
Ql 0
Q3
The O-junction m e a n s simply
Q1 + Q2 + Q3 = 0
E = same in all three b o n d s .
A T e e - j u n c t i o n in a h y d r a u l i c line is a typical s i t u a t i o n r e q u i r i n g use o f an

O - j u n c t i o n in a p o w e r b o n d graph,
T
Q1 * P ~ Q3
61
It is obvious that at the Tee-junction, pressure is common. Thus we have
P P
Q1 0 Q3
which means
Q1 + Q2 + Q3 = 0
p = same in all three bonds
to represent what is happening at the Tee.
The other summing junction, represented by 1 and called a I-JUNCTION, recognizes

algebraic summation of effort variables at conunon flow, and is typically
E2 Q
E1 E3
1
Q Q
which means
E1 + E2 + E3 = 0
Q = same in all three bonds.
A typical use of the 1-junction occurs when we wish to allow for a frictional
resisting force opposing the motion of a hydraulic cylinder-driven load, as follows:
~--~-- Xm
Fa ~ I Im I
J Ff
7/////////
where Fa is pressure induced force acting on mass Im to induce displacement Xm ;
Ff is friction force.
The power bond graph representation,

Rf
Fa Fm
1
is i n t e n d e d to e x p r e s s
Fa - Ff = Fm ,
the net force a c t i n g o n the mass. It is a p p a r e n t enough that the p i s t o n rod, the
mass, and the friction force all e x p e r i e n c e the same v e l o c i t y , leading to the
c o m p l e t e d 1-junction,
Rf
Fa Fm
1
w h i c h means, algebraically,
F a + F f + Fm = 0
Xm = same for all b o n d s
Ff = Rf(~) or vice versa.
The m e t h o d o f d e t e r m i n i n g the signs of the terms b e i n g added at summing junctions

w i l l b e i l l u s t r a t e d shortly.
5.2.10 S u m m a r y o f B a s i c Terms and S y m b o l s
S e c t i o n s 5.2.1 to 5.2.8 cover the b a s i c terms and symbols needed to form a

structure o f a p o w e r b o n d graph m o d e l of a h y d r a u l i c control system.
Fig. 5.1 shows the terms and symbols in s u m m a r y form.
5.3 F O R M I N G P O W E R BOND G R A P H S T R U C T U R E S
B e f o r e p r o c e e d i n g , let us c o n s o l i d a t e use of the b a s i c b o n d graph terms and

symbols b y d e v e l o p i n g b o n d graph s t r u c t u r e s for several w e l l k n o w n h a r d w a r e
components.
5.3.1 I n e r t i a Load with F r i c t i o n
It is c o m m o n that the load d r i v e n b y a h y d r a u l i c control system is in this

category. Fig. 5.2 shows the d e v e l o p m e n t o f a b o n d g r a p h for an i n e r t i a load w i t h
friction. Some p o i n t s to n o t e are:
Step (a): the load system has to receive p o w e r as a force v e l o c i t y product.
Step (b): the absorption b y friction o f some o f the a c t u a t i n g p o w e r is recognized

83
TERM SYMBOL IN B O N D GRAPH MEANING AND COMMENT

.......
Power Bond P o w e r flow i n the b o n d equals the
product of E and Q .
Effort E Effort means potential.
Flow Q Flow means flowrate.
SE
SE E f f o r t is constant, or independent.
Source
Q
E
SQ SQ Flow is c o n s t a n t , or i n d e p e n d e n t .
E1 E2 E1 t r a n s f o r m s to E 2 , Q1
TF - TF
Q1 Q2 t r a n s f o r m s to Q2 , such t h a t
Transformer EI-QI = E2.Q2 .
E1 E2 E1 transforms to Q2 , Q1
GY ~ GY-
Q1 Q2 t r a n s f o r m s to E2 , such t h a t
E1.QI = E2-Q2 .
Resistive Effect R R Recognizes a resistive effect.
Capacitive Effect C - - C Recognizes a capacitive effect.
Inertive E f f e c t Recognizes an i n e r t i v e or i n d u c t i v e
I I
effect.
E E
0-Junction
0 A l g e b r a i c s u m m i n g of flow v a r i a b l e s
a t c o n s t a n t e f f o r t , i.e.
Q1 + Q2 + Q3 = 0 , E = c o n s t a n t .
EIQ2
E1 E3
1-Junction Algebraic summing of effort variables
Q
a t c o n s t a n t flow, i.e.
E1 + E2 + E3 = 0 , Q = c o n s t a n t .
E21Q
FIG. 5.1 SUMMARY OF BASIC BOND GRAPH TERMS, SYMBOLS, OPERATIONS
ACTION BOND GRAPH DEVELOPMENT

(a) P o w e r a p p l i e d to load as Fa
force velocity.
Rf
(b) F r i c t i o n force r e c o g n i z e d
by Rf at 1 - j u n c t i o n . Ffi~
j ~Xm Fm l e f t as n e t force. Fm
Fa
(e) Fm
Ff Im
a c t s on load i n e r t i a
(mass) .
compl e te d.
Structure l
I
!
Fm Im
Xm
(d) Power product symbols

added to e a c h bond.
Ffl
Rf
F.~ rr I F.m Im
X X
FIG. 5.2 DEVELOPMENT OF BOND GRAPH FOR INERTIA LOAD WITH FRICTION
84
by the resistive effect Rf , at a 1-junction to show that it is a

force summation. Net force Fm emerges.
Step (c): Fm acts to accelerate the load inertia; this is recognized b y placing
Im at the end of the Fm bond.
Step (d): the two power symbols appropriate to each bond are included on the
structure.
Formation of equations from the bond graph will be dealt with later in the Text.
5.3.2 Hydraulic Cylinder
The power cylinder transforms hydraulic power (pressure flowrate) to

mechanical power (force velocity of ram) . For its dynamic model, it is decided
to include friction (piston head and gland seals) and the capacitive effect of the
cylinder end volume. For the present, the effect on dynamic response of back
pressure on the low-pressure side of the piston will be neglected.
Fig. 5.3 illustrates steps in the development of a bond graph structure.

Points to note are:
Step (a) : it is recognized that the cylinder will receive power as the product of
pressure and flowrate of oil, P Q .
Step (b) : it is recognized that the cylinder will provide power as the product of
force Fa and velocity Xa . The implication is that the cylinder
w4ZZ be connected to a resisting load.
Step (c) : it is considered that the capacitive effect of the cylinder chamber
receiving flow is dynamically significant. A certain (small) portion
Qc of the inflowing oil is absorbed in compression of the oil in the
cylinder. Thus, the effect requires summation at an 0-junction and
is designated as a capacitive e f f e c t Ca .
step (d) : pressure P transforms to force F on the piston, and flowrate Qa

to piston velocity xa . The area A of the piston is the trans-
formation modulus.
step (e): the presence of a frictional force Ffa opposing motion of the ram,
and due to piston seals and rod glands, is recognized and allowed for
as the Rfa bond from a 1-junction.
Step (f): the inertia la of the piston and rod will absorb some power as it is
accelerated. This is allowed for by the Ia bond; it leads from a
85
mPQ Xa
I "
J
Fa
(a) Power received as P

pressure - flowrate. Q
(b) Power output will be as Fa
force velocity. ~a
(c) The capacitive effect of Ca
cylinder volume is
recognized, as the Ca PIQc
bond from an 0-junction.
PO
"~
P
ea
(d) Hydraulic power is

transformed to mechanical
power. Piston area A
is the transformation
modulus.
~0 ~," A ~a
Rfa
(e} Friction is a l l ~ e d for by
Rfa at a 1-junction. 1 FfalXa
~a
(f) Inertia of ram is allowed
I i
for by Ia at a
1-junction. This
completes the structure.
0~~~1 ~1
Ca Rfa Ia
(g) Power variables added to
PIQc FfalXa F21Xa
each bond.
....... P ....
Q
0 ~ P
Qa
~
A
"-~1 "-1Fo
~a Xa ~a
Ca
(h) Simplification if cylinder
inertia Ia and friction PI Qc
Rfa effects are neglected.
P 0 P Fa
FIG. 5.3 D E V E L O P M E N T OF A B O N D G R A P H S T R U C T U R E
FOR A HYDRAULIC CYLINDER
1-junction as it is force (effort variable) F2 which is being allowed
for. The designated effects have all been accounted for. An output
bond is joined to the last 1-junction to indicate the actuator will be
joined to a load of some kind. ~his completes the structure.
Step (g) : Appropriate power product variables are added to each bond.
The bond graph structure developed for the cylinder is reasonably detailed. It
can be simplified by neglecting the cylinder rod inertia and friction (they can often
be lumped with the inertia a n d friction of the load attached to the rod). Such
simplification means neglect of the Rfa and Ia effects on Fig. 5.3g in which case
the bond graph structure becomes Fig. 5.3h
5.3.3 Induction Electric Motor
Before a dynamic model is formed, or as it is being formed, decisions have to be

made as to what effects are to be included in the model. For the present example,
we decide that motor (speed) slip, inertia of the rotor and shaft assembly, friction
in the bearings and windage in the rotor, and compliance of the motor shaft all affect
the dynamic performance of the induction motor.
Fig. 5.4 shows the steps in development of a bond graph structure, starting with
a constant voltage (SE) electric power supply, and ending with a torgue (To) and
speed ( ~ o ) output power product. Points to note are:
Step (a): supply voltage is recognized as constant (Source SE ).
Step (b): it is recognized that the motor gives out power as the mechanical
product torque (To) speed (co) .
Step (c)~ the gyration (GY) of constant voltage to constant motor field speed
(Source S~) is recognized.
Step (d): the speed slip essential to the action of an induction motor is
recognized by the P~ bond from an O-junction. It is a resistive
effect.
Step (e): torgue loss due to friction is allowed for by the resistive Rf bond
leading from a 1-junction.
Step (f): absorption of some torgue to accelerate the motor rotor and shaft
assembly is allowed for by the Ix bond leading from a 1-junction.
Step (g): the torsional compliance of the output shaft is allowed for with the
Ck bond leading from an 0-junction, recognizing that some speed
67
(a) The motor draws

power from constant i
voltage supply.
(b) It gives out power SE To

as torque and speed i ~o
at its shaft end.
(c) Electric power
transformed to
SE GY Ti To
mechanical power.
i S~ ~o
It is gyration, as
SE induces S~ ,
R~
the synchronous
field speed. Ti TilA~
is ideal torque.
(d) Speed loss due to
slip. Re describes
GY
Ti
S~
0 Ti To
~o
slip characteristic.
i~ Rf
(e) Torque lost to
friction, in
bearings, etc.
Rf recognizes
To
friction. G Y m
0 ~~ 1 ' ~~x ~o
R~ Rf Ix
(f) Torque absorbed
accelerating motor-
shaft inertia Ix .
GY ~ 0
I { +
,, 1 ,a, 1
To
OJX
To
~00
R~ Rf IX Ck
(g) Speed variation due
to compliance of
motor-shaft,
This completes
Ck .
I { I +
structure. SE GY
~ 0 ~ I I To 0 To
l
Rf Ix Ck
(h) Structure with
power state
variable symbols
added.
Ti law Tf ] ~x
SE
l
GY Ti
S~
0 Ti
~x
I 1 ~To 0 -~
(i) Simplified
structure, if shaft Rm
compliance Ck ,
rotor inertia Ix , TO lAw
rotor friction Rf
are all neglected,
and motor field
~-----0
S~
To
~o
speed ~ is
regarded as constant
(i.e. as a source
S~ ).
FIG. 5.4 DEVELOPMENT OF B O N D G R A P H F O R I N D U C T I O N ELECTRIC MOTOR
RECOGNIZING S L I P (R~), F R I C T I O N (Rf), I N E R T I A (Ia), A N D
SHAFT COMPLIANCE (Ck).
68
fluctuation along the shaft is probable in dynamic operation. The

bond structure is now complete, the outlmat bond To mo from Step (b)
fitting to the Ck 0-junction. The implication is that the To - ~o
bond will be joined to the component or system which the motor will
drive, for example a hydraulic pump.
Step (h): dual variables appropriate to each bond have been added. These are
the variables which will appear in the dynamic equations.
It can be seen that the bond graph comprises a main power flow sequence from
left to right consisting of power bonds, transformer, 0-junctions and 1-junctions.
From the main stream, power bonds lead from the junctions to the various R , C and
I effects judged to affect the motor dynamics.
As far as its effects on a hydraulic system which it might be driving are

concerned, the bond graph of Fig. 5.4 is quite detailed. It is likely that for many
applications, some of the included effects are insignificant. For example, consider
the situation when a hydraulic system is being driven by an electric motor of
adequate power, and the analyst's main interest is to predict the dynamic response
of the driven inertia load. He will probably decide that the electric motor's
inertia, its friction, and its shaft compliance all have negligible effects on load
motion. On Fig. 5.4, this means eliminating the Rf , Ix , and Ck bonds and
their functions. (which are meaningless without the bonds). In addition, the analyst
is unlikely to be interested in describing the gyration between voltage and motor
field speed, and would start his structure from the constant field speed as a source.
In these circumstances, the detailed bond graph of Fig. 5.4h can be reduced to
Fig. 5.4i. The only dynamic effect contribhted by the electric motor under these
circ~nstances is its speed slip characteristic.
5.3.4 A Simple System
Consider the hydraulic cylinder driven load combination shown on Fig. 5.5a.
The bond graph structures for each of the two major components, previously developed,
are shown in Fig. 5.5b and c, directly beneath the appropriate hardware components.
Actuator inertia and friction are neglected, knowing that they can be lumped with the
load's values in this rigidly connected situation.
The system bond graph structure is formed by simple joining of the component
bond graphs as shown in Fig. 5.5d. The Fa - ~ bond directly represents the
cylinder rod which connects cylinder piston to load. The ease with which the
component structures are joined to form the system structure is a characteristic of
the power bond graph technique.
89
a. The Sy s tern
Xm
Pa i >
Qs
W f ~ r - f r f
Friction
b. Cylinder Bond Graph (Inertia,
Friction, and Return Line c. Load Bond Graph
Effects Neglected)
, pa ~ pa TF Fa Fa I Fm
es ~ Qa A Xm ~m ~m
PalQc FflXm
ca Rf
d. System Bond Graph, Formed by Joining b and c at their Common Bond,

Fa ~tm
Pa 0 i Pa TF Fa I Fm Im
Qs Qa A Xm
Pa ! Qc FflXm
Ca Rf
e. System Bond Graph with Power Flow Direction and Causality Assigned
(Ref. Section 5.4.6)
~ Im
Ca Rf
FIG. 5.5 BOND GRAPH FOR SIMPLE CYLINDER-LOAD SYSTEM

70
More detailed bond graph structures of the linear actuator driven load will be
developed in Chapters 7 and 8. Later structures will allow for return line effects,
compliance in the actuator-load connection, internal cylinder leakage, and close-
centre, open-centre, and tandem-centre valve control.
Fig. 5.5e shows the bond graph with power flow directions (half arrow on each
bond) and causality (short transverse bar across one end of each bond) decided. The
meanings of these decisions will be presented next.
5.4 POWER FLOW DIRECTIONS, AND CAUSALITY
5.4.1 Introduction
So far in Chapter 5, we have described the basic power bond graph terms and
symbols, and have used them to form structures for some components or sub-systems of
the kinds included in typical hydraulic control systems. Later, we shall develop
more sub-system structures, and also structures for complete systems. A power bond
graph structure recognizes the presence of the various R , C and I effects which
dictate the dynamic response of the system. The structure does not indicate the
relationships by which the dynamic effects are described. That comes later.
However, the structure is useful only if it is leading towards eventual preparation
of a set of dynamic equations which describe the system dynamics. The assigning of
power flow directions, and of causality, are steps preparing the bond graph to yield
the appropriate set of equations. Assigr~ent of power flow directions provides the
signs of the terms in the algebraic s~um~ing equations to be eventually written for
each 0-junction and 1-junction. Assignment of causality reflects decisions on which
power state variable is to be cause, and which is to be effect, on each R , C and
I bond and its equation.
5.4.2 Directions of Power Flow
As with all dynamic modelling procedures, a sign convention must be adopted so

that the signs of terms in e_ouations are compatible with each other and with system
action. As with other flow diagrams, a conventional flow direction is indicated by
arrowheads. This does not imply that flow is always in the designated direction.
Some flows may be positive or negative at different times during the dynamic action.
However, if the signs of the equation set are correct for one (any) instant, they
will be correct globally, i.e. for the entire action. Adoption of a sign convention,
and sticking to it, acts to ensure that the signs of terms in the model equations are
correct. Common sense should be utilized in the designation of a sign convention - -
the convention should reflect real system behaviour as far as is possible.
A realistic approach to getting signs correct is:

71
imagine the system to be in equilibrium, at time t = 0 ,
imagine the system to be activated at t = O ,
decide signs b y the physical action which takes place in the system
immediately after it is activated.
With bond graphs, it is power which is flowing in each bond. It becomes

necessary to indicate the direction of power flow in each bond. The convention
adopted is that power is shown as entering each R , C and I element by a half
arrowhead as follows:
R C I
The directions of power flow in the remaining bonds of a bond graph follows physical
common sense - - p o w e r flows from the source to the load, usually from left to right
across the bond graph.
Consider the bond graph structure developed for the hydraulic cylinder in
Fig. 5.3, and shown again on Fig. 5.6. Obviously, it is intended that power flows
from left to right, i.e. from the hydraulic supply P Q towards the driven load.
Hence, directional half arrows are added on the m a i n (horizontal) power flow path
from left to right. For Ca , Rf and Ia , show the half arrows pointing into the
effect, utilizing the convention that power flows into these elements.
It is usually a simple matter to designate the power flow direction on each bond.
For the present, this m e r e l y helps in appreciating what is going on in the system.
Later w e shall use power flow directions in deciding the signs in the equations being
formed from the structure.
Technically, designation of power flow in a certain direction reflects a

decision that the flow variable is directed in that direction at the instant considered.
But keep in mind that we are concerned with dynamic effects, i.e. changes from
equilibrium states rather than steady-state values of variables. A flowrate can drop
from one positive value to another lower positive value; the dynamic change is
negative, b u t the physical flow remains positive:
5.4.3 Causality
As discussed in Chapter 4, consideration of whether a power state variable is a

cause variable or is an effect variable gives insight into the dynamic nature and
behaviour of the system. The idea of causality --i.e. the designation of a power
72
Ca Rf Ia
Fa L~
~a
FIG. 5.6 BOND GRAPH OF H Y D R A U L I C C Y L I N D E R (EX. FIG. 5.3),

WITH POWER FLOW DIRECTIONS ADDED
Ca Rf Ia
i ~0
I 1 T
FIG. 5.7 BOND G R A P H OF H Y D R A U L I C CYLINDER (EX. FIG. 5.6),
WITH CAUSAL BARS ADDED.
73
variable as a cause can be extended and applied to a power bond graph structure
so as to show how the equations to b e prepared for R , C and I effects are to be
arranged. Basically, it is m o s t appropriate to write effect = f(cause) ; for
subsequent digital simulation, it is preferred that a differential relationship be
expressed in integral rather than derivative form.
A causal decision is indicated b y a short transverse bar, called the causaZ bar,
drawn across one end of a power bond. For example,
with the causal bar drawn at the R end of the power bond, indicates that the R
expression is to be written in the form
Q = R(E)
where R() indicates "a resistive function of (what is in parenthesis)". That is,
it has b e e n decided that E is the cause and Q is the effect for this particular
R element.
If the causal bar is placed on the power bond at the end away from R ,
this is a statement that the R equation is to be prepared in the form
E = R(Q)
which says that Q is the cause, and E the r e s ~ t i n g effect.
In short, a causal decision on an R , C or I power bond indicates which of

the dual power variables o n the bond is to be used to compute the other. The
object of the equation ~ i.e. the left-hand side in this Text - - is the effect
variable, the subject (the variable in the right-hand or functional side of the
equation) is the cause.
Causal assignment should reflect reality both as regards to how the system
components behave and h o w particular equations should be arranged. For example,
computer solution (simulation) procedures prefer equations in integration forms
rather than in differential forms. Thus, capacitive equations should be expressed:
74
w i t h the causal b a r a w a y from the C end o f the p o w e r bond, l e a d i n g to e q u a t i o n s

such as
E = C(Q) = (possibly) I/C IQ d t + E o
instead of the differential form
Q = C(E) = (possibly) C
w h i c h w o u l d b e i m p l i e d if the c a u s a l b a r w a s a t the C end. Note that C has been

u s e d in the above e x p r e s s i o n s to i n d i c a t e a c a p a c i t i v e effect, a capacitive function
of, and in the p o s s i b l e e q u a t i o n s as a constant.
O n the o t h e r hand, inertive e q u a t i o n s should be expressed:
w i t h the causal b a r at the I end, l e a d i n g to N e w t o n ' s law b e i n g e x p r e s s e d as
Q = I(E) = i/I JE dt + E(O) (or ~ = i/Im F n e t dt + Xm(O) )
rather than i n the d i f f e r e n t i a l form
(or Fnet = Im X m )
w h i c h is implied if the causal b a r is away from the I .
For the r e a s o n s given, C and I elements on a p o w e r b o n d g r a p h are normally

g i v e n p r e f e r r e d causalities, the causal b a r s b e i n g a w a y from a C e l e m e n t , and
a d j a c e n t to an I element. R e l e m e n t s can have e i t h e r causality, d e p e n d i n g upon
the p a r t i c u l a r R and on causalities flowing from the p r e f e r r e d C and I decisions.
The total causal scheme for a system m u s t m a k e sense. One d e c i s i o n will affect
others.
C a u s a l i t y c a n be i n t e r p r e t e d that the e f f o r t v a r i a b l e i s d i r e c t e d at t ~ cca~sal

b~r, a n d consequently, t h a t the flow v a r i a b l e is d i r e c t e d azJay from the causal bar.
For example,
75
Im
means that force F is directed at the mass Ira and velocity X is a resulting
(back) effect (which is not to say that velocity is backwards, or negative:).
Conversely, for a hydraulic accumulator,
means that flowrate Q is directed into the accumulator C and pressure P is a

resulting (back) effect. That is, pressure P is directed at the causal bar, away
from C . This physically sensible interpretation helps in completion of the causal
decisions for a system, as it dictates the causal bar state at each of the summing
junctions (0 and 1-junctions).
For example, there can be only one effort variable at an 0-junction, as previously
demonstrated, indicating that there can be only one causal bar at an 0-junction,
!
lo I
If, as in the above case, a C element occurs at an O-junction, the other bonds at
the junction ~ s t have causal bars away from the junction.
On the other hand, there can be only one flow variable at a 1-junction, as
previously demonstrated, and this means that there, can be only one bond at a 1-junction
without a causal bar at the junction. Representation of an I element demonstrates
this
--Ill
with the preferred causal bar against the I end of the I bond, the remaining two
power bonds nTust have causal bars at the 1-junction if the symbolic representation is
to make physical sense.
Finally, assigning causality is a quite separate operation from assigning power

76
flow directions.
5.4.4 Example: Hydraulic Cylinder
As a more complete example, consider the bond graph structure for the hydraulic
cylinder from Section 5.3.2 and Figs. 5.3 and 5.6. Fig. 5.7 shows the bond graph
with causality added, the required steps being:
i. Preferred Causalities: place causal bar away from Ca , and against Ia .
2. Consequent Causalities: at the Ca 0-junction, the two horizontal bonds ~ s t

now have causal bars away from the junction. Add them.
At the Ia 1-junction, both horizontal bonds ~ s t now have causal bars at

their junction ends.
3. Remaining Causalities: at the Rf 1-junction, both the Rf and TF bonds

~st now have causal bars at the 1-junction.
Thus, simply using the C and I preferred causalities, the remaining

causalities fell into place. It is not always so unambiguous, but more of that
later.
5.4.5 Example : Induction Electric Motor
Fig. 5.4 showed the development of a bond graph structure for an induction
electric motor. Fig. 5.8a shows it with power flow directions assigned. Fig. 5.8b
shows the bond graph with causality assigned. Considering Fig. 5.8a,
i. Assign preferred causalities to Ck and Ix , by placing a causal bar at the

O-junction end of the Ck bond, and at the Ix end of the Ix bond.
2. The causal bars of the other two bonds at the Ck 0-junction must now be
placed at the bond ends away from the 0-junction.
3. The causal bar of the left bond at the Ix 1-junction now must be at the
1-junction.
4. Because the right-hand bond at the Rf 1-junction cannot now have a causal
bar at the junction, the remaining two bonds must have causal bars at the
junction.
This leaves the gyrator bonds and the Rm 0-junction as causally undecided,
except that the right-hand bond from the 0-junction cannot have a causal bar
at its junction end. Rational thought lets us resolve this.
77
a. Power Flow Directions Decided
K~ Rf Ix Ck
SE
i
~ GY
'~0
] "-1
] ] ~1 "0
1 To
~o
b. Causality Decided
P~ Rf Ix Ck
sE ~,
i GY I
,
1171
0 "-', 1 ", 1 : ~ 0 :o .:
FIG. 5.8 BOND GRAPH STRUCTURE FOR INDUCTION ELECTRIC MOTOR

(EX. F I G . 5.4)
78
5. A fundamental characteristic of an induction motor is that load causes speed

slip, and slip induces torque to drive the load. That is, as far as the
torque-slip characteristic is concerned, slip is the cause and torque is the
resulting effect. Therefore we should express their relationship as
T = R(~)
by placing the Re bond's causal bar at its 0-junction end. This immediately
locates the causal bar for the adjacent left bond at its GY end, leaving only
the SE/i bond causality undecided.
6. Voltage SE is a constant, unaffected b y the system. It is also the effort

variable. Effort variables are directed at the causal bar. Thus SE must
be directed into the system and its causal bar should be placed at the GY
end of its bond.
The causal assigr.nent is now complete, and the bond graph is ready to have the
dynamic equation set prepared from it.
5.4.6 Example: Simple Cylinder-Load System
Consider the system of previous Fig. 5.5 whose bond graph structure was shown on
Fig. 5.5d. Using only the criteria
power flows into R , C , I elements,

power flows from source to load,
causal bar at the I end of I bonds,
causal bar at the 0-junction end of the C bonds,
only one causal bar at an 0-junction,
causal bars on all bonds except one at 1-junctions,
the power flow directions and causal bars of Fig. 5.5e are simple to complete.
5.4.7 Summary
The reasons and the techniques for assigning power flow directions and causality
have b e e n presented and discussed. Their significance will be confirmed in Section
5.5, where equations will be developed from power bond graph structures. However,
patterns which reflect b o t h computational requirements and physical reality have
emerged, and will be summarised here.
Power flow direction and causality decisions are intended to help in the
processing of a rational set of dynamic equations from a power bond graph
structure.
79
Given a bond graph structure, allot each C and I element its preferred
causality; c effects have the causal bar placed away from the C ; I
effects have the causal bar placed against the I .
Assign appropriate causality to any designated sources.
Use the physical facts that
an 0-junction can have only one causal bar adjacent to it,
a 1-junction can have only one bond without a causal bar adjacent to the
junction,
to decide causalities consequent upon the C , I , and S causality

decisions.
If all remaining causalities do not flow automatically from the above steps,
seek the most physically obvious causal decision(s) to allow completion of
causal assignments.
Each power bond must have a causal bar, at one end only.
If causal conflicts arise, they must be resolved. If they cannot be resolved,

there is something wrong with, or missing from, the power bond graph structure.
This is a key check against validity of the proposed model. It does not
necessarily follow that a bond graph without causal conflict is complete.
As with all dynamic modelling procedures, experience and engineering judgement
enhance the probability of success.
It is best to decide causalities on the complete system bond graph, rather

than on each component bond graph before they are put together. Causality
of a component is not necessarily unique, and may have to be re-thought w h e n
the component goes into a specific system. Indeed, the bond graph structure
of a component may have to be re-thought when it is to be used in a particular
system.
Reflection and experience will confirm that at a TF power transformer there

can be only one causal bar
ie or {
but not ~ TF I nor 1
For GY power transformation the opposite holds, i.e. there must be either
80
SYMBOL MEANING AND COMMENT
Power Bond P o w e r flow in the b o n d e q u a l s the

product of E and Q .
Effort Effort means potential.
~ow FIow m e a n s flowrate.
SE E f f o r t is constant, or i n d e p e n d e n t
SE
Source
Q o f the system.
E F l o w is constant, or i n d e p e n d e n t of
sQ
SQ the system.
E1 E2 E1 t r a n s f o r m s to E2 , and Q1
TF -TF
Q1 Q2 t r a n s f o r m s t o Q2 such that
Transformer EI oQI = E 2 - Q 2 .
E1 GY E2 E1 t r a n s f o r m s to Q2 , and Q1
GY
Q1 Q2 t r a n s f o r m s to E2 such that
E I . Q I = E2-Q2 .
Resistive E f f e c t R e c o g n i z e s a resistive effect.

Capacitive Effect R e c o g n i z e s a capacitive effect.
Inertive Effect R e c o g n i z e s an inertive o r inductive
effect.
O-Junction
0 7 E0__ Q3 A l g e b r a i c s u m m i n @ o f flow variables
at same E , i.e. Q1 + Q2 + Q3 = 0 .
E [Q2
1-Junction E1 I E3 A l g e b r a i c s u m m i n g o f e f f o r t variables
- -Q -7 at same Q , i.e. E1 + E2 + E3 = 0 .
E2JQ
P o w e r F l o w Directioni s h o w s d i r e c t i o n o f p o w e r flow i n each
(half arrow bond.
o n one e n d
E a s i l Y d e c i d e d for m a i n p o w e r flow
o f each
se que nce.
bond)
P o w e r flows into R , C , I
e l e m e n t s (a convention).
G i v e s s i g n c o n v e n t i o n for equations
to b e formed for junctlons.
iCausality ~ ! Shows w h i c h of the two v a r i a b l e s on a

I ---------I b o n d is cause a n d w h i c h is effect.
(short b a r or
Decides how R , C , I equations
across one
e n d o f each ~ are to be arranged.
p o w e r bond) O n l y one causal b a r can b e a t an
0-junction. O n l y one b o n d can be at
a 1 - j u n c t i o n w i t h o u t a causal bar.
C e l e m e n t s have causal b a r away from
the C end. I b o n d s h a v e causal
b a r a t the I end.
E f f o r t is d i r e c t e d a t the causal bar,
flow away frcm it.
FIG. 5.9 SUMMARY OF COMMON BOND GRAPH TERMS AND SYMBOLS
81
two or zero causal b a r s at the GY
ie - - ~ I or I GY--
b u t not I GY - - I nor I GY l - -
Fig. 5.9 shows t a b u l a t e d the c o m m o n b o n d graph terms and symbols. Fig. 5.9 is
similar to Fig. 5.1, b u t w i t h p o w e r flow d i r e c t i o n a n d causality included.
5.5 PREPARING EQUATION SET
5.5.1 Introduction
The m a j o r objective of forming a p o w e r b o n d g r a p h and a s s i g n i n g p o w e r flow

directions and c a u s a l i t y to it is to p r o v i d e steps towards the formation o f a set of
equations w h o s e solution p r o v i d e s the p r e d i c t e d r e s p o n s e of the system.
Steps for forming the e q u a t i o n s from a b o n d g r a p h w i l l now be demonstrated, and

then s u m m a r i s e d into a r e a s o n a b l y formal procedure.
5.5.2 Example: A Simple C o m p o n e n t
C o n s i d e r the b o n d g r a p h for the i n e r t i a load w i t h f r i c t i o n s h o w n on Fig. 5.10

(Fig. 5.2, w i t h p o w e r flow and c a u s a l i t y assigned). S t a r t i n g w i t h the d o m i n a t i n g
inertia effect, the r e l a t i o n s h i p s i m p l i e d on the b o n d g r a p h are:
i. = I(Fm) --- a r r a n g e m e n t d i c t a t e d by causal b a r at Im .
= i/Im [Fin dt + ~co) --- N e w t o n ' s l a w p r o v i d e s the actual equation.

J
2. Ff = R(Xm) a r r a n g e m e n t d i c t a t e d by causal bar b e i n g

away from Im .
Kf " ~n if v i s c o u s f r i c t i o n assumed; Kf is
co e ffi ci ent.
sgn ~ F f u + Kf ~ --- i f coulomb friction Ffu is included.
3. Fa = ? to be s p e c i f i e d if F a is an e x t e r n a l input,
or to b e d e s c r i b e d from b o n d g r a p h structure
o f the c o m p o n e n t d r i v i n g the load, as
d i s c u s s e d i n the n e x t example.
4. Fm = Fa- Ff s u m m i n g junction equation, F m b e i n g net

force on load. Signs s a t i s f y p o w e r flow
h a l f arrows, i.e. Fa - Ff - Fm = 0 ; Fm
chosen as o b j e c t (left-hand side) o f
e q u a t i o n as F a and F f are a l r e a d y
o b j e c t s o f equations.
Equations 1 to 4 are the model, formally d e r i v e d from the b o n d graph. Note t h a t

82
Xm
Im
Fa I/2////z
Ff
Rf
FIG. 5.10 BOND GRAPH FOR INERTIA LOAD WI~ FRICTION
a. System
xm
Pa
m
Qs
r
Im
Friction
b. Power Bond Graph
l Pa Pa ~ TF Im
0 Qa " A Fa ~ 1 1 ~ x m
xm
Ca Rf
FIG. 5.11 BOND GRAPH FOR CYLINDER DRIVEN LOAD SYSTEM

83
there are four p o w e r state v a r i a b l e s a s s o c i a t e d w i t h the b o n d graph (Fa , Xm , Ff ,

Fm) and that e a c h i s the o b j e c t o f a p o w e r state equation. It is p r o b a b l e t h a t l o a d
displacement Xm(t) w o u l d b e o f p r i m a r y i n t e r e s t to the analyst; it is therefore
added as an a u x i l i a r y e q u a t i o n b y describing:
5. Xm = (~ dt + Xm(0) --- simple i n t e g r a t i o n of ~ .

J
5.5.3 An E l e m e n t a r y S y s t e m E x a m p l e
Fig. 5.11 s h o w s the b o n d g r a p h for a c y l i n d e r - d r i v e n inertia load w i t h friction.

It was derived, w i t h o u t p o w e r f l o w or causality, in S e c t i o n 5.4.6 and Fig. 5.5. The
supply flow of oil to the c y l i n d e r m u s t come f r o m an u p s t r e a m c o m p o n e n t s u c h as a
4-way control valve. However, for the m o m e n t l e t us k e e p it simple. As w i t h the
previous example, w e s h a l l work b a c k from the driven load in p r e p a r i n g the equations:
I. ~ = i/Im IFm d t + ~ ( 0 ) --- Im equation, as p r e v i o u s l y .
2. Ff = Kf X m --- Rf e q u a t i o n assuming v i s c o u s friction.
3. Fm = Fa - Ff --- net force acting o n mass; e q u a t i o n o f the

1-junction.
4. Fa = Pa A --- t r a n s f o r m a t i o n o f c y l i n d e r p r e s s u r e Pa to
force Fa via m o d u l u s A .
5. Qa = A'~ --- t r a n s f o r m a t i o n b e t w e e n s w e p t volume rate of

p i s t o n and p i s t o n v e l o c i t y v i a m o d u l u s A .
6. Pa = C(Qc) --- d i c t a t e d b y causal b a r functional a r r a n g e m e n t

of the Ca equation.
= 1 / C a |Qc d t + Pa(0) --- a s s u m i n g linear c a p a c i t a n c e w i t h Ca a

J coefficient.
7. Qs = ? --- d e t e r m i n e d b y n e x t u p s t r e a m component.
S. Qc = Qs - Qa --- e q u a t i o n o f the 0-junction, signs taken from

p o w e r flow arrows; Oc chosen as o b j e c t
because Qs and Q a a l r e a d y are o b j e c t s of
equations.
The equation set 1 to 8 is the p o w e r state m o d e l for the system; e a c h p o w e r state

variable is the o b j e c t o f one equation; a u x i l i a r y e q u a t i o n s can be a d d e d if it is
desired to s t u d y o t h e r than p o w e r state v a r i a b l e s , or i f some o f the R , C , I
quantities are o t h e r than simple c o e f f i c i e n t s a n d require further description. For
example, e x c e p t for small p e r t u r b a t i o n s a b o u t a m e a n value Xm , the c a p a c i t a n c e o f
the h y d r a u l i c c y l i n d e r is n o t a simple c o e f f i c i e n t Ca . This is b e c a u s e the v o l u m e
of the cylinder end, and hence o f the oil u n d e r p r e s s u r e Pa , c h a n g e s s i g n i f i c a n t l y
with Xm . Remember, a C , R , and e v e n I symbol on a b o n d g r a p h does n o t
84
necessarily refer to a coefficient, although very o f t e n this is the case. More

fundamentally, a C refers to "a capacitive function", R to "a resistive function,
and I to "an inertial function". A coefficient is, of course, the simplest
functional relator between two quantities.
It is informative and significant that the causal bar had to be placed at the
input end of the Pa Qs bond. It conveys that the component directing flow to
the actuator is essentially a f!owrate provider, and that pressure is a consequent
back e f f e c t dictated b y the actuator and load system. The bond graph reveals the
fallacy of thinking that pressure can be supplied to an actuator.
5.5.4 A More Complete System
System and Assumption8

Fig. 5.12a shows the system. It is desired to p r e d i c t l o a d m o t i o n xm(t)
following control commands Xv(t) to the valve spool.
For the present analysis, it will be assumed
that the pumping system supplies oil to the control valve at constant pressure
and that pump supply is adequate. This allows neglect of the dynamic effects
of all components upstream of the control valve, reducing the hardware to be
considered to that shown on Fig. 5.12b. Further, supply pressure to the
valve is designated as a source, SP ,
that the flow in the return lines (cylinder to valve to tank) has a
negligible effect on load motion response Xm following valve actuation,
that the resistance o f the h i g h pressure side line is negligible.
The assumptions allow reduction of the load and cylinder bond graphs to those, used
previously, shown on Fig. 5.12c under the respective components of Fig. 5.12b. We
have not previously required a b o n d graph for a 4-way valve.
4-Way Control Valve

The closed-centre 4-way (i.e. 4 port) valve h a s four control orifices each of
which can be regarded as a resistance to flow. It is control of these resistances
b y the operator which provides control of the system. In the central position for
an ideal closed-centre valve, each resistance is infinitely high, and each port fl0w
is zero. As the valve spool is moved, control orifice resistances decrease with
increasing Xv , falling to minimum values w h e n the ports are fully open. Control
valve flows, including open-centre and tandem-centre arrangements, are discussed in
detail in Chapter 7.
85
When return line effects are neglected, only the supply port and the pressure-
side control port need be considered. Further, the ports are physically close, and
together with the internal passage connecting them, they can be lumped as a single
resistance. It is of course pressure that is "lost" during flow through a restriction.
Thus, for dynamic analysis purposes, the left-hand bond graph structure of Fig. 5.12c
summarises the contribution of the valve to the present system's dynamic action. What
we have is
constant supply pressure SP and its associated supply flowrate Qs on the

input bond. Note that Qs is not c o n s t a n t - - i t is both pressure-drop and
valve-displacement dependent,
valve resistance accounted for as the Rv bond, leading from a 1-junction to

show that pressure drop ~Pv (effort variable) is being accounted for,
pressure Pa (= SP - APv) and Qs go forward from the valve to the next

connected component.
System Bond Graph

Fig. 5.12d shows the bond graph for the three connected components, including
power flow directions and causal bars. Using the conventions of causality, only the
causal bar locations on the SP Qs and the Rv bonds were in doubt. At the left
1-junction, prior satisfaction of C and I causalities leaves the causal bar for
Pa Qs at the 1-junction end, as follows,
Rv
APvIQs
SP Pa
1
Qs Qs
with causality for the other two bonds to be decided. To resolve this, consider the
physical reality of either bond. For the Rv bond, for example, it is usual to
express flowrate through the valve as a function of pressure drop
Qs = f(APv, Xv)
Thus, the causal bar on the Rv bond needs to be at the bond's Rv end. This means
in turn that the causal bar on the SP bond must be at its 1-junction end.
Alternatively, if supply pressure is a source SP , it cannot be a back effect

(load dependent). Using the idea that the effort variable is directed at the causal
bar, the SP bond causal bar must be in the forward power flow direction, against
the 1-junction. ~nis decision, together with the already established causal bar for
86
a. The Sy stem
r- -I
xm
iI I, I II I, t ii ///////"
b. The Dynamic Components if Supply Pressure at Control

Valve Assumed Constant Xm
I Xv r
sp
i...................I ill f f ff
c. Component Bond Graphs, if Return Line Effects Neglected and

Xv is either Positive or Negative Only
Rv Rf
APviQs
sp I Pa 0 Pa Fa Fa I Fm , Im
Qs Qs ~
Pa IQc
Ca
d. System Bond Graph
Rv Rf
Apv~Qs
Qs Qa A Xm Xm
Ca
FIG. 5.12 BOND GRAPH FOR A SYSTEM

87
the P a - Qs bond, m e a n s that the Rv b o n d causal b a r m u s t be a t its Rv end.
Note h o w the need to resolve a d o u b t f u l causality gives i n s i g h t into the s y s t e m ' s

c a u s e - e f f e c t behaviour.
System's Equations
There are ten d i f f e r e n t p o w e r variables o n Fig. 5.12d, and so we need ten p o w e r
state equations. We shall p r e p a r e them in the o r d e r
Sources
I effects
C effects
R e ffects
TF
Junctions (0 to 1 from l e f t to right), giving,
Source : SP = constant
Im: = I/Ira [Fm a t + ~(0)
Ca: Pa = 1/Ca IQc d t + Pa(0)
Rf: Ff = R(Xm) . To b e specified~ e.g.
= Kf - Xm for v i s c o u s friction, Kf being a coefficient
Rv: Qs = R(APv) from causal d e c i s i o n
= K X v - sgn APv APv (see C h a p t e r 7 for d e t a i l e d

explanation)
TF: Fa
= Pa A [ a r r a n g e d to h a v e p r e v i o u s l y u n u s e d o b j e c t
Qa = Xm - A J variables
left 1-junction: APv = SP - Pa

a r r a n g e d to have p r e v i o u s l y unused o b j e c t
0-junction: Qc = Qs - Qa variables, and u s i n g p o w e r f l o w d i r e c t i o n
a r r o w s to decide signs
r i g h t 1-junction: Fm = Fa - Ff
as the set of p o w e r state equations. A u x i l i a r y e q u a t i o n s w h i c h will be r e q u i r e d are:
Xv = (input to be specified)
Xm = I~ dt (desired p r i m a r y r e s p o n s e variable).
J
In a s p e c i f i c system, the various i n d e p e n d e n t p a r a m e t e r s and c o e f f i c i e n t s n e e d

88
to be specified numerically (SP , Im , Ca , Kf , K , A , Xv) whereon the

equations can be simulated to provide response predictions. Some examples shall be
carried through to complete solutions in Chapter 8.
5.5.5 Summary
The examples given can be worked back to include a pressure relief valve,
hydraulic pump, an electric motor and other components. However, this will be left
to future examples and case studies. The purpose at present is to illustrate how
the equation set is formed from a b o n d graph structure.
There is not an exclusige order of the equations. However, the bond graph
novice is advised to seek equations in the following sequence. Having obtained a
bond graph with power flow directions and causality assigned,
write an equation for each source,
write at least the functional relationship for each R , C , I , with

functional arrangement dictated by causality,
write the effort equation for each TF and GY , at least in functional form,
using as the object a variable not yet the object of an equation,
write the flow equation for each TF and GY , again using as the object a
previously unused variable,
write a summing equation for each 0 and 1-junction, using power flow
directions to dictate signs, and choosing as the object of each equation a
power variable not already the object of an equation,
check that there is an equation for each power state variable, and that no
such variable is without an equation,
decide any specific relationships in the p o w e r state equations as yet

undecided,
write an equation for each of any system variables required for study which
are not among the power state variables,
write defining equations for any parameters appearing, b u t as yet undefined,

among the power state equations. Typically, a control input m i g h t exist
(such as control valve displacement X v ).
To get a solution from the model,
assign numbers to all coefficients appearing in the completed equation set,

89
simulate (solve) to get specific system's response.
With some experience gained, an analyst m a y prefer to prepare the equations from
a bond graph in a somewhat different order; typically to start at the first (or last')
summing junction. For the systems of Fig. 5.12d, this would result in
First (left) 1-junction: SP = constant
QS = R(APv) = K Xv sgn Pv Pv ~
APv = SP - Pa
First (left) 0-junction: Pa = 1/Ca IQc dt + Pa(0)
Qc = Qs - Qa
and so on. With this approach, the choice of object for each summing equation is
less certain, and may require subsequent re-arrangement of the equations.
The power bond graph technique results in a set of relatively simple equations,
readily checked for errors, and readily programmed for computer solution. The same
forms of equation appear in example after example, a n d those relationships relevant
to hydraulic control systems quickly become ingrained and are easily recalled and
used.
It is possible to prepare from bond graphs equations in the conventional transfer

function and state-variable forms. Karnopp and Rosenberg [5.1] describe in detail
how to develop the state-variable form X = A X + B U . This development will
n o t be presented nor encouraged in the present Text as it appears to offer no
advantage. Indeed, the structural rigidity required for transfer function and
state-variable model forms may be a positive hindrance. Suffice it to keep in mind
that models in conventional forms can be formed from bond graphs if there is a
compelling reason.
5.6 SOME FURTHER ASPECTS OF BOND GRAPHS
Bond graph methods well beyond w h a t appears in the present Text have been
developed and used. The present Text concentrates only on those aspects of bond
graph methods which the author has found to be useful and appropriate to the dynamic
analysis and performance prediction of industrial hydraulic control systems. Readers
wishing to delve further into bond graph techniques can do so via References 5.1 to
5.4. However, there are several aspects of b o n d graphs not discussed in detail in
the present Text which should be b r o u g h t to the attention of readers.
90
5.6.1 Modulation of Effects
In some systems, one or more effects may be modulated during system action.
The modulation can be independent of the system, or can be made via automatic feed-
back. In either case, it is usual that the modulation action does not utilize
significant system power. It is usual that one system variable, rather than a power
pair, is utilized to effect any internal system feedback. For example if a variable
displacement pump was adjusted manually during operation, no system power is utilized.
If a solenoid-operated hydraulic valve is actuated b y a system pressure, no power is
drawn from the hydraulic system itself. In general, logic, control, and information
signals are sensed, processed, and conveyed w i t h o u t loss of power from the powered
control system either because they are v e r y low-powered signals, or because external
power is utilized.
There are several ways modulation signals, if they exist, can be highlighted on
bond graphs. For example, in Fig. 5.12 it is intended that the control valve be
manually operated such that its resistance Rv is Xv dependent. This can be
illustrated as
R v (Xv) Rv --4~-Xv
o a T
If modulation derives from external sources, it is convenient to use the left-hand

representation. If modulation derives from feedback within the system, the right-
hand method allows its physical origins to be illustrated.
For example, consider a pilot-operated pressure relief valve where the relief
valve is situated at the pump but its o p e r a t i n g pressure is sensed well downstream
at the actuator. Fig. 5.13a illustrates this situation, and includes the pilot
actuation line as a b r o k e n line. Fig. 5.13b shows the corresponding part of a bond
graph representation of the system, and includes the modulating feedback effect as a
broken line w i t h a full arrowhead on it. Thus, the modulation bond is distinguished
from power bonds b y being a b r o k e n line of light construction, and b y having a full
arrowhead on it in the direction of signal flow. Note that the modulation bond,
referred to in bond graph terminology as an a~tiue or a~tivatsd bond, has only one
power variable associated with it, in this case pressure Pa . The implication is
that the other power variable is zero and therefore that no power flows in the active
bond.
5.6.2 Modulated Transformers
The modulus of a TF or GY effect sometimes may be variable. The variability

91
a. The S y s t e m
P~p, Relief Valve, Filter, ~ n e 4-Way Valve Hydraulic Motor Load

/ \/ x/ ......... \ /--~-N
@
I
I
u_ Pilot L i n e J
b. Bond Graph
Pilot Signal
J %
/ %
/
I
Rvl %% Cal
!
Rr
/
/
1 ",1
~--,- Vp
~,-.,~ 0 ~-.,. 0 ,--- 0 1 ~-~ ~
Pa2 ~ / /
1
Rp
T
Cp Ka
J
Rv2 Ca2
~.,, / \ / k I \ /
Pump, Relief Valve, Filter, Line 4-Way Valve Hydraulic Motor Load
FIG. 5.13 SHOWING FEEDBACK ON A BOND GRAPH
Rr represents the Relief valve.

Cal represents the inlet side of the
Hydraulic Motor.
(The components of the bond graph will
be explained in Ch. 7).
may be deliberately controllable by an external input, or it may be controlled by

system feedback. A variable delivery pump is an example. TE or GY elements in
this category are called modulated transformer8 and can be represented on bond graphs
as MTF or MGY . Further, an active bond can be shown to indicate the source of
the modulus variation. For example
92
I I1~ MTFJ
i i
Vp
i l l u s t r a t e s h o w a p r e s s u r e - c o m p e n s a t e d p u m p c o u l d be indicated. The symbol

i l l u s t r a t e s simply that p u m p d i s p l a c e m e n t p e r radian Vp is a function o f p r e s s u r e
P , w h i c h could be a l o c a l l y s e t or a fed-back pressure.
5.6.3 F i e l d s and J u n c t i o n S t r u c t u r e s
There is a more s o p h i s t i c a t e d a p p r o a c h to b o n d graphs i n v o l v i n g field8 of c ,

I , and R effects, w i t h junction 8tln~ctul~98 o f 0-junctions, 1-junctions, TF's
and GY's to link the fields. F i e l d s are m u l t i p o r t g e n e r a l i z a t i o n s w h i c h allow
concise r e p r e s e n t a t i o n on b o n d g r a p h s of complex a s s e m b l a g e s of C's , I's , and
R's . There are situations i n v o l v i n g d y n a m i c a l l y complex m e c h a n i c a l m o t i o n s or
linkages, or complex e l e c t r i c a l circuits, in w h i c h the field and junction structure
c o n c e p t is advantageous. However, it appears t h a t the d i r e c t e f f e c t - b y - e f f e c t
approach to b o n d graphs is quite appropriate to the m o d e l l i n g o f m o s t h y d r a u l i c
control systems. It is l e f t to the reader to take up study of fields and junction
s t r u c t u r e s v i a Ref. 5.1 w h e n he is r e a d y or in need.
5.6.4 Simplifications
w i t h a little usage, it w i l l become a p p a r e n t that a d j a c e n t similar junctions

can be combined. For example, on p r e v i o u s Fig. 5.8, the Rf and Ia 1-junctions
are a d j a c e n t and can be e x p r e s s e d
Rf Rf
I b~
I
Ia Ia
In the b o n d graphs of this Chapter, b o t h v a r i a b l e s of e a c h b o n d w e r e e v e n t u a l l y

a d d e d to the diagrams. A t e a c h junction, there is n e e d to show the e f f o r t v a r i a b l e
(0-junctions) or the flow v a r i a b l e (1-junctions) on only one of the b o n d s at the
junction, as it is common o n all b o n d s at the junction. It is quite appropriate to
s h o w the common v a r i a b l e a g a i n s t the junction itself, as follows
93
I Q2 E2I
0 ~ or EZ ~ I E3___E!__L
Q1 E Q3 Q
5.7 CONCLUSION
Bond graph terms and symbols have been introduced and their uses demonstrated.
Only those aspects of bond graph techniques which have been found to be relevant to
the present Text have been discussed in detail. For further reading in bond graph
techniques and their uses in a wide variety of dynamic systems, Refs. 5.1 to 5.4
are recc~unended.
It is reccm~nended that the user be systematic when putting power state

variables on bonds. In the present Text, the effort variable is written above
horizontally drawn bonds, and to the left of vertical bonds. The flow variables
are written below the horizontal bond and to the right of the vertical bond.
It is recommended also that, initially at least, readers adopt the formal

approaches to assigning causality and to preparation of equation sets given in
Sections 5.4 and 5.5 respectively.
Chapter 6 gives a brief review of digital simulation, the model solution

method most appropriate to bond graph models. Chapter 7 introduces in some detail
the equation forms which might be required for use with the bond graph of a
hydraulic control system. Chapter 8 demonstrates the formation and use of bond
graph models for a variety of hydraulic systems, including electrohydraulic servo-
systems with feedback shown as active bonds.
5.8 REFERENCES
5.1 Karnopp, D. and FDsenberg, R., System Dynamics; a Unified Approach,

Wiley, 1975.
5.2 Thoma, J., Introduction to Bond Graphs and their Applications, Pergamon,
1975.
5.3 Transactions ASME J. of Dyn. Syst. Meas. and Control, V.94, Series G,
N.3, 1972 (whole issue).
5.4 Journal of the Franklin Institute, V.308, N.3, 1979 (whole issue).
CHAPTER 6
SOLUTION OF POWER FLOW MODELS
6.1 INTRODUCTION
Dynamic mathematical models of physical engineering systems of the kinds relevant

to the present Text, are formed in order to be used to predict a system's dynamic
response, or to provide dynamic performance information. It is intended that the
model be solved, fully or partially, to provide this information. When choosing a
modelling procedure, the analyst needs to be aware of the solution procedures
available to him. If, as in the not too distant past, only manual solution procedures
are available, simplified models amenable to hand solution must be sought. If an
analogue computer is to be utilized, a more complete model can be sought. However,
there will be limitations on the number and types of non-linear expressions which can
be included in the model. Digital computers provide the potential for solution of
fully detailed dynamic models. In particul~r, the digital simulation procedures
developed for use with most of the major computers provide a general solution
procedure for a very wide range of dynamic models.
Power flow modelling procedures provide the system model as a set of low-order
algebraic and differential equations. A detailed model of a system of some substance
could be a set containing a substantial number of such equations. Digital simulation
procedures are most compatible with this situation. There is no n e e d - - i n fact it
is undesirable - - to run the low-order equation groups together to form more complex
groupings of variables and coefficients. In the system design situation, in which
the sensitivity of system performance to changes of some parameters is likely to be
investigated, the simple presentation of coefficients and variables resulting from
power flow modelling is positively an advantage.
9S
Thus, power flow modelling and digital simulation are highly compatible. The
existence of the digital simulation solution procedure has made it possible to realize
the advantages of the power flow modelling procedure. Digital simulation is the only
solution method considered to be relevant to the present Text.
Digital simulation will be dealt with only superficially. It is required here

only as a user-orientated model solution procedure particularly suited to the purposes
of the present Text. Each major computer is likely to have its own digital simulation
procedure(s) or variations. Experience has shown that it is relatively easy to learn
to utilize them. Though internally complex, the procedures are designed to be
utilized by the user in a "black box" mode. The user has only to learn how to get
his data, equations, and instructions into the "black box" solution procedure. In
fact, the basic purpose of a digital simulation procedure is to bridge the user to the
computer so that the user's simple statements can trigger and control the computer's
complex and comprehensive problem-solving capability.
A number of relevant references are listed at the end of Chapter 6.
6.2 DIGITAL SIMULATION
The digital simulation procedures relevant to the present Text are generally
classified as Continuous System Simulation Languages ICSSL's). A continuous system
is regarded as one which can have a continuous time history associated with its
dynamic performance, rather than one intended or known to have discrete states in
time.
Fig. 6.1 illustrates the relationship of the CSSL's among the dynamic problem
solving languages con~nonly used in digital computing. To help orientation of the
reader, some of the specific languages associated with each grouping are included.
The high level problem solving languages such as Fortran can be utilized directly to
prepare a model for solution. However, the CSSL's (for continuous systems) and the
discrete system languages provide the bridge between the user and the computer's
basic compiling language (Fortran or alternative).
The CSSL's fall into t~o categories. Those in the block-orientated group are
intended to fairly directly rival analogue ccmputing, in which a block diagram type
of model is most appropriate. The expression-orientated CSSL's anticipate receiving
the model in the form of a set of equations. It is this group which is of primary
interest in the present Text, and only this group will be considered further.
Characteristics desirable in a CSSL include:

96
DIGITAL
SIMULATION
High Level General

Discrete System
Problem Solving
Simulation Procedures
Languages
FORTRAN GASP
ALGOL GERT
COBOL SIMULA
BASIC etc.
etc.
Continuous System
Simulation Procedures
(CSSL's)
Expression-Orientated Block-Orientated
CSSL's CSSLfs
MIMIC DYNAMO
CSMP MIDAS
CSSL PACTOLUS
ACSL etc.
etc.
FIG. 6.1 APPROACHES TO DIGITAL SIMULATION

easy utilization, with little computing expertise required of the user;
good man-machine rapport;
good result display capability;
economy in computer use;
speed, in both utilization and computing;
accuracy: numerical instability is a potential problem. To avoid it requires

choice of an integration procedure, and integration step size, compatible with
the particular problem being solved. For efficient use of the computer,
multiple rate is preferred. This allows integration step size to vary from
integration variable to integration variable, and even during a particular
integration. A slowly changing variable does not require the small
integration step size necessary with rapidly changing variables;
parallel statement of model equations: this allows model sections or

equations to be worked out and specified in any order. The computer does
the sorting into its execution (i.e. procedural) order. Problem languages
such as Fortran require procedural programming, i.e. the computer executes
the equations and statements in the order they are supplied. Though parallel
programming is an illusion .... the computer must convert the statements
received by it into procedural form before execution starts - - it is real as
far as the programmer is concerned. Parallel statement specification is much
preferred to the procedural form for most sections of a digital simulation;
the ability to specify sections of a model in algebraic form, with separate

specification of numerical values for coefficients and parameters. This
macro capability is particularly desirable in system design, when the effects
of value changes need to be studied. It is also very useful where response
optimisation is required;
the ability generally to incorporate common operations, functions, and signals

as "building blocks" which can be called into required positions in a system
being simulated. In general, this macro capability allows access to stored
"library" sub-routines (preferably in either Fortran or the CSSL language).
It should also include a capability for logic control of statement execution
- - this is needed to represent some common non-linear relationships.
6.3 EXPRESS ION-ORIENTATED CSSL' S
The CSSL's date back to the mid-1950's. Ref. 6.1 contains a survey of their
98
early developments, which were of considerable diversity. In 1967 the Simulation

Councils Incoporated (now the Society for Computer Simulation) published specifications
for standardizing CSSL development. Ref. 6.2 contains details. The most recent
CSSL's are close to b u t not precisely in line with the specifications. The
specifications are themselves ~albiguous or incomplete in parts. Improved accessing
methods and new requirements complicate the search for user-advantaging standardization
of CSSL's. Ref. 6.3 gives information on developments to the mid-1970's, and
Ref. 6.10 gives a hydraulic control system's orientated review of the situation to
1979.
MIMIC (Control Data Corporation) and DSL/90 (IBM) were perhaps the major links
drawing the early digital simulations towards the standard CSSL concept. Both were
widely and generally used. Control Data Corporation, influenced by the standardization
move, now has CSSL 3 [6.4]. Similarly, IBM's DSL/90 has been developed into CSMP 3
(Continuous System Modelling Program), [6.5]. Xerox Data Systems SL-I [6.6] is
particularly close to the standard CSSL. Mitchell and Gauthier [6.7] have produced
ACSL, pronounced "axle" and referring to Advanced Continuous Simulation Language, for
use with most computers. Most CSSL's have been produced for use with a particular
company's computers. Thus, ACSL represents a major step forward.
Apart from the general purpose CSSL's, there exist many special purpose orientated
digital simulation procedures. For example, HYDSIM [6.8] is especially designed for
hydraulic control system response analysis using power port models of components.
ENPORT [6.9] has been especially developed to accept a model in bond graph form. It
has the disadvantage (very serious in the present Text) of being able to handle only
linear expressions. Both HYDSIM and ENPORT have the significant feature of being
able to store "standard" models of components likely to be used and re-used in various
control systems. Some other CSSL's are orientated to incorporate response optimization
schemes.
6.4 CONCLUSION
CSSL procedures provide the designer-analyst of engineering control systems with

the model solution procedure closest to his needs. They can have the solution speed,
accuracy, man-to-system rapport, and the display and plotting characteristics of
analogue computers. Their computational capability far outstrips that of analogue
computers. So too does their general availability. Increasing availability of
interactive access and of minicomputers allows the "hands on" access previously the
preserve of analogue computers.
Digital simulation is not necessarily a panacea for solving dynamic models.

99
Programming and running time can be costly if inexpertly used. Round-off error and
precise accuracy may be difficult to estimate. Numerical instability can be a problem
where different parts of a model have widely different response speeds. This
situation may be reflected by the presence in the model of widely different natural
frequencies. These kinds of model are referred to as stiff. Numerical stability
in the solution of stiff models is being better overcome with the development of more
flexible integration procedures which do not necessarily require formal explicitness
of all of the relationships of the model. ACSL edges into this category. It may
require some experience in particular problem areas before the best integration
procedure and step size specification can be made with confidence. A ~odern CSSL
may have as options several Runge-Kutta integration procedures plus several predictor-
corrector types (i.e. Milne). Some are more complex than others, requiring more
computer capability. Thus a simple self-starting Runge-Kutta 4th order procedure
should be preferred to the more complex Milne 5th order form (needs starter, complex
but very accurate) unless it is known to be inadequate to a particular set of
equations or relationship.
CSSL's are very compatible with power flow models. They are also very compatible
with the equation sets and relationships which are relevant to hydraulic control system
response, including response optimization procedures.
Results of completed simulations of bond graph models using digital simulation

are included in Chapter 8.
6.5 REFERENCES
6.1 Branden, D., Burwen, M., and Wachouski, A., Digital Simulation of Physical
Systems, Assoc. for Computing Machinery.
6.2 Strauss, J.C., The SCI Continuous System Simulation Language (CSSL),
Simulation, V.9, N.6, Dec. 1967, pp.281-303.
6.3 Benyon, P.R., Improving and Standardizing Continuous Simulation Languages,

Proc. SIMSIG Simulation Conf., Melbourne, Australia, 1976.
6.4 Continuous System Simulation Language 3 (CSSL 3) Version i, Control Data

Corp., Sunnyvale, California, 1974.
6.5 Continuous System Modelling Program 3 (CSMP 3) and Graphic Feature, IBM
Canada, Ontario, Canada, 1972.
6.6 SL-I Reference Manual, Xerox Data Systems, E1 Sequndo, California, 1970.
6.7 Advanced Continuous Simulation Language (ACSL) User Guide/Reference Manual,

Mitchell and Gauthier Assoc., Concord, Mass., 1975.
6.8 Smith, C.K., HYDSIM User's Manual, School of Engineering, Oklahoma State
University, 1973.
100
6.9 Rosenberg, R., A Users Guide to ENPORT 4, Wiley Interscience, 1975.
6 .i0 Iyengar, S.K.R., Review of Computer Software for Simulating Hydraulic

Systems, BFPR Journal, V.12, N.4, pp.323-329, Fluid Power Research Centre,
Oklahoma State University, 1979.
6 .ii Benyon, P.R., Some Recent International Activities in Simulation, Proc.

4th Biennial Conf. Sim. Soc. of Aust., University of Queensland, Australia,
1980.
CHAPTER 7
SELECTING EQUATIONS AND COEFFICIENTS
7.1 INTRODUCTION
The bond graph material of the Text to here has emphasized formation of the
structure of a system's dynamic model rather than the actual equations whose solution
will provide dyn~m~ic response. Equations have b e e n presented in some examples, b u t
with little discussion.
It is valid to emphasize the model structure. To the experienced bond graph

analyst the model structure is the model. He "sees" his physical system in the bond
graph structure. In a particular technology .............in
... the present case hydraulic
control systems .... the physical equations used are of relatively few forms and these
appear over and over again in the models of different systems w i t h i n that technology.
Thus the b o n d graph analyst also "sees" the equations implicit on the structure. To
be sure, there is often a choice to be made on how detailed a description (i.e. an
equation) of a particular action should be in a particular analysis. The represent-
ation of friction is one case. The representation of valve flowrate versus pressure
drop and valve displacement is another.
However, specific equations for each S , R , C , I , TF and GY on a bond

graph do have to be specified before simulation can progress. Further, w h e n all
equations have been selected and written down, there remains the selection of
numerical values for the various coefficients in the equations. For a particular
system whose hardware has b e e n provisionally specified, many of the coefficients
required for simulation of the dynamic model can be obtained as design specifications
or from component manufacturer's data. However, there is usually a small number of
coefficients whose values are less easily determined.
I02
The present Chapter brings together physical relationships which are relevant to
the Text. Where ambiguity or choice exists, some discussion is offered. Selection
of values for those coefficients which are not straightforward is discussed, and some
values for coefficients are included.
7.2 COMPLIANCE
7.2.1 Introduction
No liquid is fully incompressible. No mechanical connection is completely

rigid. Both retain elasticity characteristics which can be likened to the character-
istics of stiff springs. The compliance characteristic of the oil in a hydraulic
control system is a vital parameter affecting system dynamic response. Mechanical
compliance in the various mechanical components of the system may affect system
response, depending on the degree of compliance present. Hydraulic lines - -
particularly hoses - - can expand radially under pressure, mechanical shafts can
suffer torsional strains, actuator rods suffer compression or elongation, beams
suffer deflection, and so on. Because of its significance to system performance,
oil compliance will be considered in some detail. Mechanical compliance, being
better understood by most engineers, will be dealt with briefly only.
7.2.2 Oil compliance
Oil compliance is a key characteristic in the action and dynamic performance of

hydraulic control systems. Initially it might be considered desirable that the oil
has a very low compliance (i.e. very high resistance to compression), an incompressible
liquid being the ideal. Reflection indicates that this is not the case. An
incompressible liquid subjected to positive pumping action which is opposed by a
significant downstream impedance, as is the case with most hydraulic control systems,
would experience instant and extreme high pressure development. This would impose
intolerable strain on the mechanical components containing the oil (pump lines,
filters, valves, actuators). It takes time for even the fastest-acting pressure
relief valves to respond. On the other hand, it is obvious enough that a highly
compressible (i.e. high compliance) fluid is incompatible with fast response, as the
stiff coupling between control and load necessary for fast response is lost. A
pneumatic servo can never be as stiff and as fast in response as its hydraulic
counterpart.
Fig. 7.1 illustrates the idea that for a particular hydraulic system and its
driven load, the pressure development at the actuator following a control command at
the control valve might look like
curve 1 if the oil was virtually incompressible

(i.e. very low compliance, or very high bulk modulus)
103
Pressure
i. Ultra-High Bulk Modulus
l i
f {
II Ii f~% /r-- 3. Correct Bulk Modulus
' t / ..~.
# i ; i .i., ,
; i 'it i '. ,," /'.." ""--'-"

; I ,/ t i ~,, / -
#' ~ f/ t #- -" -./
; ~ li ,-
i /___..l
Time
FIG. 7.1 ILLUSTRATING THE EFFECT WHICH BULK MODULUS

OF F L U I D C A N H A V E O N T H E P R E S S U R E DEVELOPMENT
IN A HYDRAULIC CONTROL SYSTEM
Pl
,2 I i I Im
FIG. 7.2 FLOW THROUGH VALVE INTO ACTUATOR

104
curve 2 if the oil w a s quite c o m p r e s s i b l e

(i.e. h i g h compliance, o r v e r y low b u l k modulus)
curve 3 if the oil c o m p l i a n c e w a s a b o u t right.
Thus, the actual v a l u e o f b u l k m o d u l u s for a h y d r a u l i c s y s t e m l i q u i d is important.

It s h o u l d n o t b e so h i g h t h a t e x c e s s i v e p r e s s u r e s are b u i l t up, nor so low that stiff
d i r e c t control of the load is lost. It seems that l i g h t m i n e r a l oils, w i t h a natural

q u i e s c e n t b u l k m o d u l u s of a r o u n d 1750 M P a are a b o u t right. It is n o t clear whether
this is close to a n a t u r a l o p t i m u m figure, o r w h e t h e r h y d r a u l i c c o m p o n e n t s and systems
h a v e e v o l v e d r e c o g n i z i n g t h a t m i n e r a l o i l s w e r e their n a t u r a l fluids.
The c o m p l i a n c e c h a r a c t e r i s t i c of a liquid can be d e s c r i b e d b y a b u l k m o d u l u s

d e f i n e d b y the static e q u a t i o n
B = - V AP/AV 7.1(a)
where V is the initial v o l u m e of l i q u i d s u b j e c t e d to c o m p r e s s i o n

AV is the change in v o l u m e V due to c o m p r e s s i o n
AP is the change in p r e s s u r e w h i c h causes c o m p r e s s i o n AV .
If v o l u m e V of l i q u i d is s u b j e c t e d to ~ n c r e a s e d p r e s s u r e AP , V dew,Base8 by
A V , i.e. AP/AV is negative, and the n e g a t i v e sign o f e q u a t i o n 7.1(a) is u s e d to
make bulk modulus B a p o s i t i v e value.
For d y n a m i c p e r f o r m a n c e analysis, the need is to h a v e a r e l a t i o n s h i p w h i c h

e n a b l e s c a l c u l a t i o n o f p r e s s u r e d e v e l o p m e n t in c h a m b e r s into w h i c h h y d r a u l i c fluid is
flowing. Fig. 7.2 shows a t y p i c a l case flowrate ol at pressure Pl upstream
of the o p e n control v a l v e b e c o m e s 02 at P2 as it flows into the actuator.
N e g l e c t i n g leakage losses, t r a n s i e n t Q2 w i l l be less than Q1 b y an a m o u n t
Qc k n o w n as the c o m p r e s s i b i l i t y c o m p o n e n t o f the i n c o m i n g flowrate. T h a t is,
Q2 = Q1 - Q c d e s c r i b e s the d y n a m i c flows f o l l o w i n g a c t i v a t i o n of the c o n t r o l valve.
In s t e a d y - s t a t e Q 1 = Q2 , as the e f f e c t o n the s p e c i f i c v o l u m e o f the f l u i d o f
the low s t e a d y - s t a t e p r e s s u r e d i f f e r e n c e can b e neglected.
E q u a t i o n 7.1(a) c a n be e x p r e s s e d
B = - V dP/dV 7.1(b)
This e q u a t i o n can b e m a d e d y n a m i c b y u s i n g dP/dt for dP and dV/dt for dV ,
giving
B = - V (dP/dt)/(dV/dt) = - V P/V 7.1(c)
Now for the situation o f Fia. 7.2. V = -Oc , and

10S
B = V P/Qc 7.1(d)
~ene
P = B / V JQc dt + P(o) 7.1(e)
E q u a t i o n 7.1(e) is readily seen to b e a l i n e a r c a p a c i t a n c e type relationship,

with c a p a c i t a n c e d e f i n e d as
C = V/B 7.1(f)
E q u a t i o n 7.1(a) can be e x p r e s s e d
B
AP . . . . AV 7.1(g)
V
For a p a r t i c u l a r B , V and AV , AP w i l l be a p a r t i c u l a r value. For the same

B and AV , b u t an i n c r e a s e d volume V , AP will b e smaller. That is, p r e s s u r e
development is d e p e n d e n t o n the m a g n i t u d e o f the v o l u m e V under compression or
expansion. It is i m p o r t a n t to k e e p this in m i n d w h e n u t i l i z i n g e q u a t i o n 7.1(e) for
systems i n v o l v i n g linear actuators. A s oil flows into the actuator, d i s p l a c i n g its
piston, the volume o f oil u n d e r c o m p r e s s i o n changes. This is r e a d i l y a c c o u n t e d for
by e x p r e s s i n g the v o l u m e as
V = v(o) + Xa Aa 7.2
where V(o) is a suitable initial or m e a n v a l u e for the volume of oil u n d e r

cc~npression
Xa is p i s t o n d i s p l a c e m e n t from the p o s i t i o n V = V(o)
Aa is the effective area o f the actuator piston.
Note that for single rod actuators, Aa is one v a l u e for the h e a d side a n d a n o t h e r
(smaller) v a l u e for the rod side. The p r o b l e m o f v a r y i n g volume does not arise with
rotary drives, the volume of oil u n d e r c o m p r e s s i o n o n e i t h e r side o f the r o t a r y
actuator r e m a i n i n g v i r t u a l l y constant.
Oil c o m p l i a n c e m a y h a v e to be c o n s i d e r e d in any situation w h e r e oil d i s c h a r g i n g

into or through a volume is t h o u g h t to have a s i g n i f i c a n t e f f e c t o n system response
by affecting p r e s s u r e development. P o s s i b l e s i t u a t i o n s of this type include a p u m p
discharging into a line, f l o w into and t h r o u g h filters, f l o w through lines, flow into
actuators, flow from a c t u a t o r s through lines and p o s s i b l y filters o f s i g n i f i c a n t
fluid capacitance. W h e t h e r o r not the f l u i d c a p a c i t a n c e e f f e c t of a p a r t i c u l a r
component should be considered in the d y n a m i c a n a l y s e s of a system is l a r g e l y a
matter of e x p e r i e n c e and judgement.
106
7.2.3 Values for Bulk Modulus
The bulk modulus of a particular liquid is affected by its temperature and its
pressure, though the effects are not significant in the working pressure and
temperature ranges associated normally with hydraulic control systems fluids.
Aeration of the fluid, prevalent in working hydraulic systems, is a much more
significant factor. A typical mineral hydraulic fluid in its quiescent state has a
bulk modulus of around 1750 MPa. Under the influence of turbulence and cavitation in
working hydraulic systems, bulk modulus might drop locally to 10% of this value, or
even to zero locally if cavitation is fully developed. The lack of drive stiffness
caused by low bulk modulus can seriously affect the response and performance of
hydraulic control systems. In its cycle through a particular system the oil's bulk
modulus can vary from a high value in the reservoir, to a lower value after pumping,
to a lower value again after expansion through the control valve, and so on. In a
well designed reservoir, the oil has a chance to regain or approach its quiescent
bulk modulus before again being drawn into the working system. Merritt [7.1]
discusses and provides typical values for bulk modulus at different parts of working
hydraulic systems. A detailed discussion of bulk modulus and the variations in its
definition is provided in Ref. 7.2.
7.2.4 Mechanical Compliance
Hydraulic control systems include numerous mechanical components, each of which

has compliance in the form of elasticity. For example
the shaft between drive and pump is subjected to torsional deflection when
under torque,
the hydraulic lines (particularly hoses) can expand under pressure,
the rod of a loaded linear actuator is u n d e r compression or tension. It

may also be subjected to a degree of column buckling effect,
linkages used to connect actuator to load, prevalent in mobile lifting and

digging equilmnent for example, are subjected to deflection when under load.
Most hydraulic systems are designed such that the mechanical components are very
stiff and will experience minimal deflection under load. Therefore, mechanical
compliance
can often be neglected when predicting the response of a system,
if significant, is usually well within the elastic range of the component,

allowing use of well-established mechanical load-deflection relationships.
107
If it is required, mechanical compliance usually can be represented as a Ck

effect from an O-junction, as shown on Fig. 7.3. It is usual also that the Ck
effect can be taken as linear, in which case Fig. 7.3 can be interpreted as
A~ 7.3
F = i/Ck "IAx d t + F(o) 7.4
= i/Ck - AX + F(o) = Kk tuX + F(o) 7~4 (a)
where Xl , ~2 are the instant velocities of the two sections of the compliance
being considered
is the velocity difference between X1 and X2
Ck is the compliance of the elastic mechanical component
Kk = i/Ck is the "stiffness" of the elastic mechanical component
F is the force developed in and transmitted b y the compliance
AX is the instant deflection between the two sections considered.
Note that equation 7.4 is in the form required for power flow modelling whereas
equation 7.4(a) is the conventional static form for the force-deflection relationship
for a linear spring.
Equations 7.3, 7.4, 7.4(a) have the equivalents for rotation motion
= ml - ~2 7.5
T = i/Ck IA~ dt + T(o) 7.6
= i/Ck A0 + T(o) = Kk A0 + T(o) 7.6(a)
where el , ~2 are the instant angular velocities of the two sections of the
compliance being considered
is the difference between el and ~2
is the torque developed in and transmitted by the compliant
component
Ck is the compliance coefficient of the component
= 1/Ck is the "stiffness" coefficient of the component
8 = IA~ dt is the angular deflection involved.

J
If appropriate, non-linear Ck relationships can be used. For example, for

the load-deflection characteristics of non-uniform cross-section beams.
108
Ck
! F ~ 0 F ~!
~2 '
FIG. 7.3 REPRESENTING MECHANICAL COMPLIANCE
a. Viscous Friction b. Coulomb Friction

Ff Ff
ope K f T
/
X
C. Coulomb + Viscous Friotion d. Linearized Coulomb + Viscous

Ff Friction Ff
j /
/q /
FIG. 7.4 REPRESENTING FRICTION
109
7.3 FRICTION
Friction is present wherever one mechanical part moves on and relative to

another. There are many such situations within working hydraulic systems. For
example, the pistons moving in their bores in a piston pump or motor; the piston
head in a linear actuator; the piston rod moving through the gland of a linear
actuator; etc. The most significant friction, however, is likely to be associated
with the load driven by the system. Substantial loads require the support of heavy
bearings and guides.
Friction in these various circumstances can range between the two limits:
i. V{SaOU8 free,on, where good film lubrication is implied. That is, the mating
mechanical parts are separated by a film of lubricant, so that only fluid shear is
present to resist motion. Under these circumstances
Ff ~ ~ 7.V(a)
Ff = Kf " X 7.7
where Ff is the force required to overcome friction

is the relative velocity between the moving parts
Kf is the viscous friction coefficient
provides an accurate description of friction force for use in dynamic simulation.

Fig. 7.4a illustrates the Ff - X relationship of equation 7.7.
2. CouZomb J'~ei~on, where the mating mechanical components are in direct contact,
with incomplete or no effective lubrication. In this situation static friction has
to be overcome before relative motion can start, and the static friction force F~
is described by
F~ = ~ Fn 7.8
where is the coefficient of friction existing between the components

F~ is the force required to start motion
Fn is the normal force existing between the components.
It is well known that when coulomb friction is overcome and motion starts, the
friction force drops sharply. Fig. 7.4b illustrates coulomb friction.
Many of the components in a hydraulic control system are flooded with the
hydraulic fluid. One of the properties desired of a hydraulic fluid is that it be a
good lubricant. This is certainly realized if a mineral oil is used, as is the case
for most hydraulic systems. Thus, good lubrication can be expected for the internal
110
moving parts of pumps, motors, actuators, and valves. However, the presence of
generous quantities of oil does not absolutely guarantee that no direct contact of
parts takes place. This is particularly so if the parts have stationary periods,
such as is usual with linear actuators, spool valves, and hydraulic motors. A film
of lubricant developed and stabilized when parts are in motion can be squeezed out to
some degree when motion stops. It is preferable to allow for this effect when
considering friction of even well-lubricated parts, by using a friction force-velocity
characteristic of the type illustrated on Fig. 7.4c.
Fig. 7.4d illustrates a reasonable idealization of ~riction in well lubricated

components. If the dry friction band is narrow, it can be neglected in a system
simulation, leavir~ frictional force as viscous (equation 7.7). If the "stiction"
effect is considered to be significant friction can be idealized to the form
Ff = F~-sg~~f-~ 79
where sgn means "sign of", allowing that X can be positive or negative.
In extreme cases a characteristic of the form of Fig. 7.4c can be described by

equation or numerically. However, the overriding requirement for successful dynamic
analysis and simulation should be kept in mind, i.e. the simplest model adequate for
the proposed simulation is the best: It is surprising how often that an assumption
that friction in hydraulic system components is approximately viscous proves adequate
for hydraulic control system performance predictions. Of course, the more suspect
the lubrication present the more suspect is a viscous friction assumption. However,
it is reasonable that a system designer can assume that the moving parts of his
proposed system components, and of the load being driven, are reasonably lubricated.
Frictional forces and torques always oppose motion and irreversibly dissipate
(waste) power. However, it should not be presumed that friction should be eliminated
or even minimized. Friction is often the only mechanism providing damping of load
response to sharp inputs. A minimum-friction actuator driving a minimum-friction
load may well prove too oscillatory, or even result in limit cycling which borders on
instability.
Numerical values for friction coefficients are included in some of the examples
of Chapter 8.
7.4 MODELLING DRIVEN LOADS
7.4.1 Inherent Loads
The positioning of inertia loads in the presence of friction, possibly of

111
structural compliance, and possibly of external load forces, is the task faced by the
majority of hydraulic control systems. The designer-analyst needs to understand the
basic load arrangements he m a y have to choose between or adapt frc~. For example to
control the rudder of an aircraft, the control system needs to overcome inertia,
friction, structural compliance, and the aerodynamic forces on the rudder. The
hydraulic system of a large press has to cope with inertia of the press head assembly,
friction in the press guides and bearing, deflections in the press frame, and the
large load induced when the press head comes in contact with the work piece.
Fig. 7.5 illustrates eight basic load arrangement models, together with their
bond graphs, causal requirements, and probable (not all exclusive) equations. The
first three entries are straightforward. The force Fa developed in the actuator
is the causal input variable, load-actuator velocity is the resulting effect. There
is no compliance effect between actuator and load, and this is often a reasonable
assumption.
Entry 4 allows for structural compliance in the drive or mountings between

actuator and load. The C effect causes actuator velocity Xa and load velocity
to be different, by the amount ~X shown on the bond graph. The preferred
causalities for Ck and for Im both require the causal bar to be on the left end
of the Fa Xa bond. This is a significant variation from the load input bond
causalities of Entries 1 to 3, all of which implied stiff (non-compliant) connection
between actuator and load. The bond graph of Entry 4 is telling us that force Fa
is a back effect for the Entry 4 drive, and that actuator velocity Xa is the cause
effect. This is a most significant revelation, with strong implications as to how
the actuator model (bond graph) m u s t be presented for connection to a compliant drive.
Entry 5 allows for load friction in the compliant drive model.
Entry 6 illustrates a compliant drive between actuator and load, with damping or
friction also present in the connection. This might be the situation if actuator
friction was considered to be significant or even necessary to damp down oscillations
in Xa and Xm due to the presence of compliance Ck . Again, and for the same
reasons as for Entry 4, actuator force Fa is a back effect resulting from causal
velocity Xa . The 0-junction on the bond graphs tells us that the compliant drive
causes Xm and Xa to be different by the amount AX (and consequently Xm and
Xa to be different by ~X ). The 1-junction allows that both Ck and Rf affect
, and that they act in parallel with each other.
Entry 7 adds the external force Fi to the situation of Entry 6. Fi is not

necessarily constant, nor necessarily linear, and is a source.
112
Entry 8 shows the situation of Entry 3 operating in the vertical plane, when
gravitational force acting on the mass must be included. This is shown as the
constant force SFg (Source, Force, gravitationai). SFg could have been added to
the single 1-junction of Entry 3.
7.4.2 External load Forces
Most hydraulic control system loads include inertias. Many also include other
forces. For example, a machine tool drive or tool positioner will experience cutting
forces. The tail-plane control on an aircraft will experience an aerodynamic force
which is dependent on both the speed of the airflow over the surface and on the angle
of attack of the surface to that airflow. A sensitive model for an eleotrohydraulic
servovalve may have to account for both the inertia of the valve spool and the flow-
forces generated on it.
Consider the case of an aircraft's rudder control. Both the inertia of the
rudder, and the aerodynamic forces on it should be considered when assessing its
position response to input cen~ands. Whatever else may have been included in a
bond graph for the control system and its load, the structure should end with the
inclusion of an R8 , vel bond from a 1-junction and with causal bar at the junction,
indicating recognition of a force which is dependent on rudder angle (8) and on
airspeed (vel) . It is of no consequence whether Im is shown projecting from the
same 1-junction. It can project horizontally as previously, or at any other angle
for that matter.
7.5 LEAKAGE FLOWRATES
7.5.1 Relationships
The working parts of pumps, valves, and actuators must have clearances. For
example, the pistons in their bores in piston pumps and actuators, the spool in its
bore in spool valves, hhe gears and the side plates in gear pumps, etc. These
clearances, though very small, will act as leakage paths when under high pressure
drops. Normal hydraulic fluids are sufficiently viscous, and working clearances are
sufficiently small, that leakage flows can be assumed to be laminar. That is
QZ ~ dP 7.10(a)
where Q is the leakage flowrate through the leakage path

AP is the pressure drop which is inducing leakage.
Then,
Q = I/R ~P 7.10
113
EQU&TIONS C ~ v ~ HTS
1. Mall only
ga - causal f~rce Fa derlwas f z ~ dxlwi~
ece~orent.
Fa ILl
9= - iln~.[rad~+~(o) Wew~n'm law.
"J--CLt] xm - I~m d t + m . ( o ) Auxiliary equation.
2. ~all with friction Fa - causal force
l/~-I~ ~t+~(o)
R(| d l n o ~ l a " a r l l i l t i v e
function of'.
- K~,~m for visc.frlc K is visc. frlc. colff.
Fm - Fa-FZ Net fozol on I~ .
3. MIls with friction and c c m p l i ~ t

z~mrxaint mf
Fa - c~usal force
C~ Fa kl Fm ~! Im Ff " RlXm)
Fa ~ ~ ~= ,1 9= rk - c19=) CO d e n o t e s "a ~ a t i t i ~
f u n c t i o n of'.
For l i n e a r a p r i n g , o f
-I c~ff. ~k .
ok Fm = Fa-Ff-Fk
4, Ku|, with ~pllant drive ~a - c a u s a l input wa cannot be the c a u s a l

input, au development of
O~
- i/zm.lra dt+~(o) spzing deflection im
r e q u i r e d before F II
& - i,-9= ~.ilo~ed. ~hua, t a
is the causal input foz
this con flguration.
Fa = C(&X) P is a capacitive f u n c t i o n
of ~f .
FOr linear spzln 9.
5. Mass, c(~pliant drive, friction ~a - causal input As f o r 4 . Xa i s cau~ad

by t h e component d r i v i n g
= ~/z=.I~= dt+9=(o) t h e lo a d s y s t e m .
1 ~&
: ,ah " 0 ---~-~

1"a
r
~ -: m
~
ra
F~
-
-
-
~.-~
c(~)
Fa-rf
Ff RiAX) Use d e s i r e d e q u a t i o n foz
friction.
6. Mass, w i t h compliance and f~ictlon ~a - causal input AS for 4 and 5,
i n th e d r i ~ Fa ~ Fa = : I m i.e. for ~ p l l ~ t drives
xm = i/Zm-[Fa dt+Xm(o) 4, 5, 6, ~elocity is the
cause varlable and force
the resulting eZfect
variable.
Ff - R(AX)
Interesting, isn't it?
rk = c(al)
Rf ~ ~,,I~ ~ ~
Pa - Ff+Fk
~. Ks f o r 6, p l u s external f o r ~ Fi
xeslst/ng motion
~a - causal input
F - specified force SFI indicates

rt is
- c(~x)
Ill
Fm - F a F i All~Ing for affect of Fi
S. ~ f o r 3, b u t s u s p e n d e d v e r t i c a l l y
~f
Fa
~m sFq i s c o n s t a n t f o r c e
As f o r l . e x e r t e d o n ma~s due t o
rf !
Fa I gravity.
~he ~ o a d j a c e n t l - J u ~ t i ~ s
Fro' - Fa-F f - F k can be combined.
Pm - nn'-SFg
Fa (~k SFg
FIG. 7.5 INERTIA LOAD SYSTEMS, WITH VARIOUS FRICTION,

COMPLIANCE, AND EXTERNAL FORCE CONSTRAINTS
114
where R is the resistance of the leakage path.
AP will be a variable in the power state equations and the simulation. Usu~ID
leakage flows back to tank, the discharge side pressure can be taken as zero, and
approximates the pressure at the entry to the clearance. Thus, leakage from a p~mp
can be expressed
Qp = i/Rp Pp 7.10(h)
where Pp is the pressure at pump discharge

Rp is the leakage resistance coefficient of the pump.
Analytical estimation of a particular R may be complex. For example, for a

piston located coaxially in its bore, as illustrated on Fig. 7.6a, leakage is g i v e n ~
Q = ~ D H3/(96 ~ L)J"
~ Ap 7.11
where D is piston and bore diameter

H is the diametral clearance between piston and bore
is dynamic viscosity of the liquid
L is the length of the piston
AP is the axial pressure drop which is inducing leakage.
In this case, it is readily seen that for the single p i s t o n
R = 96 ~ ' L/(z D H 3) 7.12
However, the attitude of a piston working in its bore can rarely be guaranteed.
The piston can tilt, and can become eccentric in the bore as illustrated on Figs.
7.6b, c, respectively. For the case of full eccentricity, Fig. 7.6d, it is readily
shown that QZ will be 2.5 times the value calculated from equation 7.11. That is,
for full eccentricity, R drops to 0.4 of its value for the concentric case of
Fig. 7.6a. It is worth observing that leakage is proportional to clearance cubed.
That is, if clearance is doubled, the resistance is decreased to an eighth, and the
leakage flowrate goes up eight times. It is also w o r t h noting that clearance is
affected - - increased ....b y pressure expansion effects, increasing leakage from the
value calculated using manufactured clearance. Leakage in the clearances of piston
components is discussed more fully in Ref. 7.3.
7.5.2 Coefficients
Empirical evaluation of an R coefficient is relatively simple. For a

particular pump, for example, the overall Rp (pistons, port plate, etc.) can be
established by plotting measured Qp (case drain flowrate) as AP is varied in
115
a. Concentric
2. i ~," / I I ! I :
T
H/2~---- '/ / / / / f ~ / ~" / /
b. Tilted Piston
J / / J / J J / J J
/ I / / / I I'////
C. Eccentric Piston
Ll / L / / i / / / /
/ / / / / / / / / / /
d. Fully Eccentric
H q _/ / / / / / / / / / /
I
/ J / / / / ," / / / / /
FIG. 7.6 LEAKAGE PASSAGE FOR HYDRAULIC PISTONS

116
d i s c r e t e steps. The p u m p should b e running w h i l e the test is c a r r i e d out. Reli~le

pump manufacturer's data for v o l ~ e t r i c e f f i c i e n c y can also be u s e d to estimate Rp.
The same applies to o t h e r components. The e x p e r i e n c e d d e s i g n e r - a n a l y s t w i l l not have
m u c h d i f f i c u l t y in e s t a b l i s h i n g r e a s o n a b l e R v a l u e s for use in e q u a t i o n 7.10 and in
the r e l a t e d simulation.
For example, as described in Ref. 7.4, the Rp value for a 40 k W Rexroth

v a r i a b l e d i s p l a c e m e n t axial p i s t o n p u m p w a s e s t a b l i s h e d from d a t a s u p p l i e d b y the
m a n u f a c t u r e r to b e
Rip = 0.505 1012 Pa/(m3/s)
7.6 R E L I E F VALVE F L O W R A T E S
7.6 .I Relationships
A p r e s s u r e r e l i e f valve is i n t e n d e d to r e m a i n shut up to a c e r t a i n set pressure,

b e y o n d w h i c h it will o p e n e n o u g h to d r a i n o f f s u f f i c i e n t flowrate to l i m i t the
p r e s s u r e in the line it is p r o t e c t i n g to the set pressure. F o r use in b o n d graphs,
it is r e q u i r e d to have the r e l a t i o n s h i p
Qr = f(p) 7.13(a)
where Qr is the flowrate through the relief valve to tank

P is the p r e s s u r e e x p e r i e n c e d b y the r e l i e f valve.
The relief valve itself will have internal dynamics, as it is c o m p o s e d

e s s e n t i a l l y o f a d a m p e d - s p r i n g - m a s s e x c i t e d b y an e x t e r n a l force (pressure effectiv~
area) . F l o w forces will also a f f e c t the d y n a m i c s of relief v a l v e components.
However, a w e l l - d e s i g n e d r e l i e f valve will have a s p e e d of r e s p o n s e m a n y times faster
than the r e s p o n s e speed c a p a b i l i t y o f the s y s t e m ' s d r i v e n load [7.1]. That is, the
natural frequency of the r e l i e f valve will b e far h i g h e r t h a n that of the driven 10ad
system. B e a r i n g in m i n d that the system d e s i g n e r - a n a l y s t ' s p r i m a r y c o n c e r n will be
in p r e d i c t i n g load response to c o m m a n d inputs, the d y n a m i c s of the r e l i e f valve can
b e n e g l e c t e d u n l e s s a s p e c i f i c study is b e i n g m a d e of the relief valve itself.
Fig. 7.7 i l l u s t r a t e s the form o f the s t a t i c Qr(P) r e l a t i o n s h i p to be expected

from a r e l i e f valve. The r e l a t i o n s h i p can b e d e s c r i b e d b y the d i s c o n t i n u o u s two-part
equation
Qr = 0 if P .< P s e t
7.13
= i/Rr (P - Pset) if P > Pset
where i/Rr is the slope ~Q/AP of the Qr(P) line for P > Pset .
117
slope 1/Rr
P
Q1
~
PIQr
~%
U
P
Q2
Qr = f (P)
Q2 = Q1 - Qr
! ~-P
Pset
FIG. 7.7 PRES SURE-FLOWRATE FIG. 7.8 REPRESENTING A
CHARACTERISTIC OF A PRESSURE RELIEF VALVE
PRESSURE RELIEF VALVE BY BOND GRAPH
P~ Rf Ix Ck
FIG. 7.9 BOND GRAPH FOR AN ELECTRICAL INDUCTION MOTOR

(REF. FIG. 5.4)
T~
ax. Operating Torque
i| OperatingRegion
!~ Shaft Speed, ~x
Speed at Max. Operating Torque ~ I
Synchronous Fie'Id
Speed, S~
FIG. 7.10 TORQUE-SPEED CHARACTERISTIC FOR AN ELECTRICAL

INDUCTION MOTOR
118
The characteristic of Fig. 7.7 is readily established by experiment. The d e s i g n e r
analyst can expect that a relief valve manufacturer can supply it. The P > Pset
characteristic is not always a perfect straight line. It may have a curve to it, in
which case the curve can be approximated by a straight line, described b y a non-line~
relationship, or represented by data points. Alternatively, experimental data for
Qr(P) can be used in a simulation. Ref. 7.5 describes in more detail the case for
representing relief valves by equation 7.13 in system simulations. It includes the
establishing of Rr values for a number of different valves, both direct-acting and
pilot-operated. It describes also the results of simulations using the equation 7.13
representation of relief valves, and their support by experiment.
In the occasional event that an analyst may wish to study the dynamics of the
relief valve, for example to study chatter, a number of references are available
(Ref. 7.6 for example).
In stm~ary, for general system analysis a pressure relief valve can be represen~d
as an R-effect leading from an 0-junction in a bond graph, as illustrated on Fig. 7.~
with the Rr effect to be described by equation 7.13. Note that the single causal
bar allowed at the 0-junction can be o n either of the main power flow bonds, d e p e n ~
on the particular system the relief valve is used in.
7.6.2 Coefficients
As an example of Rx values, the system of Ref. 7.5 used six different relief
valves, all rated in the region 4 to 20 MPa set pressure, and 60 to 180 /min. Values
for Rr ranged from approximately 0.5 to 5 x 109 pa/(m3/s) depending upon the
particular valve and the particular set pressure. Values for Rr should be available
from the relief valve manufacturer, or should be assessable from his valve's perform~ce
data.
7.7 ELECTRIC INDUCTION M O T O R
7.7 .I Relationships
Many hydraulic pumps are driven by induction electric motors. Often the motor's
output speed c a n be regarded as c o n s t a n t , in w h i c h case the motor's internal
characteristics and dynamics can be ~eglected and the input speed to the pump can be
taken as a source (constant) . However, b y its very nature, an induction motor's
output speed varies with load. Sometimes this "speed slip" will b e significant to a
particular hydraulic system response analysis, and it is then necessary to have a
dynamic model of the induction motor. In Chapter 5, Section 5.3.3 and Fig. 5.4, a
detailed bond graph was developed for this situation. Fig. 7.9 is a reproduction of
Fig. 5.4b w i t h the gyration from source voltage to source motor field speed omitted
119
as redundant. It allows for
speed slip, as the e f f e c t Re
friction, as the e f f e c t Rf
motor inertia, as the e f f e c t Ix
output shaft compliance, as the e f f e c t Ck .
A typical induction motor slip characteristic is i l l u s t r a t e d on Fig. 7.10. Over
the operating range the c u r v e c a n b e a p p r o x i m a t e d by
T~ = R~ /~0 7.14
where T~ is the torque d e v e l o p e d o n the rotor

R~ is the slope o f t h e T~ - ~ relationship
A~ is the speed s l i p r e l a t i v e to s y n c h r o n o u s f i e l d speed S~ .
It is c o m m o n to a l l o w for f r i c t i o n and w i n d a g e i n a loaded m o t o r as
Tf = Kf ~x 3 7.15
where Tf is the torque component of T~ l o s t in o v e r c o m i n g friction
Kf is a f r i c t i o n coefficient
~x is m o t o r shaft speed
The inertia e f f e c t of the m o t o r can b e a l l o w e d for as
0~x = i/Ix ITx d t + ~x(o) 7.16
where ~x is m o t o r speed
Ix is i n e r t i a of rotor, including output shaft

Tx is that torque component of T~ which is a b s o r b e d accelerating inertia Ix
The shaft c o m p l i a n c e effect, Ck , can be d e s c r i b e d
To = I/Ck I~mk d t + To(o) 7.17
where To is the n e t torque a v a i l a b l e at the o u t p u t e n d of the s h a f t

Ck is the shaft c o m p l i a n c e (inverse o f stiffness)
A~k is the a n g u l a r speed d i f f e r e n c e , due to d e f l e c t i o n , between the m o t o r e n d
and the o u t p u t end o f the shaft.
To and ~o are r e s p e c t i v e l y the torque a n d the s p e e d a v a i l a b l e to d r i v e the
component to b e c o n n e c t e d to the motor, and the p o w e r available is To ~ o . It
should be a p p r e c i a t e d t h a t all o f the v a r i a b l e s in e q u a t i o n s 7.14 to 7.17 are dynamic.
They could all b e e x p r e s s e d as f u n c t i o n o f time - - i . e . T~(t) , /~(t) , Tf(t) ,

120
etc. a n d should a l w a y s b e v i e w e d as such.
7.7.2 Model
The c o m p l e t e m o d e l for the i n d u c t i o n m o t o r represented b y Fig. 7.9 b e c o m e s
Source : Sm = constant
R~: T~ --- R e ~
Rf: Tf = Kf mx 3
Ix: ~x = 1/Ix [Tx d t + ~x(o)
Ck: TO = i/Ck I~k d t + TO(o)

0-junction: AL0 = S0~ - ~x
1-junction: Ty = T~ - Tf
1-junction: Tx = Ty- To
0-junction: d~k = ~x - ~o
Note t h a t the m o d e l is f o r m e d on the b a s i s t h a t it w i l l be loaded; i.e. that it
will be driving some oemponent, such as a h y d r a u l i c pump, which will d e m a n d power in

the f o r m o f torque and speed.
7.7.3 Coefficients
For the model presented, the c o e f f i c i e n t s required before a simulation can

proceed are s~ , Rm , Kf , Ix , and Ck , a n d the i n i t i a l values ~x(o) and
To(o) Synchronous field speed S~ and s p e e d slip c o e f f i c i e n t Rm are basic
motor characteristics available with a particular motor. A reputable manufacturer
s h o u l d a l s o b e able to s u p p l y o n r e q u e s t the inertia, o f the m o t o r ' s r o t o r and its

output shaft, to w h i c h can be a d d e d the d e s i g n e r - a n a l y s t ' s calculation of the inertia
of any c o u p l i n g and driven component shafting present i n the drive. The friction
coefficient Kf can be a s s e s s e d e i t h e r f r o m the m a n u f a c t u r e r ' s d a t a or from mechanical
efficiency curves for the motor. Shaft compliance Ck can be c a l c u l a t e d from
conventional torsional stiffness or d e f l e c t i o n formula.
A s an example, the c h a r a c t e r i s t i c coefficients for a p a r t i c u l a r 50 kW, 240 volt,

50 Hz, 4-pole G e n e r a l Electric induction motor were found to be
Sm = 1500 r p m
Rm = 53 N m/(rad/s)
Kf = 0.316 - 10 -3 N - m / ( r a d / s ) 2
Ix = 16.5 10 -3 k g m2
Ck = i0 10 -6 r a d / ( N m)
121
7.7.4 Conclusion
The electric induction motor has been modelled in some detail. The model
developed and the particular coefficients for it will be utilized in an example in
Chapter 8 where motor dynamics were considered to be significant to the analysis.
It is emphasized, again, that it is rare that electric motor dynamics are of primary
significance in the dynamic performance analysis of hydraulic control systems.
7.8 HYDRAULIC PUMPS
7.8 .i Relationships
A hydraulic pump normally is running continuously. In dynamic analyses where

driven load response is the main objective of study it can often be assumed that pump
speed is constant. For fixed displacement pumps, inertia of the pump's moving parts
can then be neglected. So too can internal friction in the pump. That is, the
pump is regarded as experiencing a speed source, and can be represented by the bond
graph of Fig. 7.11a, where
T is the torque developed at the pump shaft

Sep is the pump shaft speed, here regarded as a constant and hence as source
Vp is the geometric displacement per radian rotation of the pump
Pp is the pressure at pump discharge
Qp is the ideal or geometric discharge flowrate of the pump, here regarded
as a constant and hence as source SQp
Rp denotes that internal pump leakage is to be considered
Qp is pump leakage flowrate
Qp is actual pump discharge flowrate.
For a particular drive speed ep and swept volume rate Vp the bond graph can
be simplified to that of Fig. 7.11b.
It is of interest and significance that the causal bar on the pump's output bond
is at the left, indicating that pressure is a back effect, to be induced by whatever
impedance the pump's discharge flowrate experiences. Thus, the bond graph indicates
unambiguously that a positive displacement pump is primarily a generator of flowrate,
and not of pressure.
There may be situations in which it is necessary or desirable to include some or

all of a fixed displacement pump's own dynamic effects in a simulation. This would
become necessary if it was thought that a pump's drive speed could vary significantly
under the oonditions relevant to a particular performance analysis. Fig. 7.11c shows
a bond graph structure which allows for pump friction (Rfp) and pump inertia (Ip) .
122
a. Fixed-Displacement, Speed b. As for a, Simplified

Con s tan t
Rp R~
Sup Vp
c As for a, but allowing for pump d. Allowing for the
friction (Rfp) and pump rotating capacitance Effect After
parts inertia (Ip) Pump Discharge
Rfp Ip Rp cp
up mp Vp Pp
e. Including Relief Valve f. Combined Rp, Cp, and Rr

Effects
Rr
~Qr
g. Detailed Bond Graph for Pump

op
Rfp
Tf~ R~p~p ~QrRr
T'
Ip cp
FIG. 7 .ii VARIOUS BOND GRAPH CONFIGURATIONS FOR

HYDRAULIC SYSTEM PUMPS
QP
t / Slightly falling, due to Qip

increasing with Pp
~ Qp falls dramatically if
Pp 9 Ppset
J| ~pp
Set Pressure
FIG. 7.12 DESIRED CF3LRACTERISTIC OF

PRESSURE-COMPENSATION OF PUMP
123
NOR that input speed can no longer be a source, as ~p has to be allowed to be

affected by Rfp and/or Ip . In general neither input torque nor input speed will
be a source in such cases. The input bond causality shown on Fig. 7.11c would have
~ be satisfied by the output bond of the device driving the pump.
For a variable displacement pump, the displacement per radian of the pump is
modulated by a suitable mechanism. For a swashplate pump for example Vp on Fig.
7.11a and c would have to be replaced by
vp = Vp' - e
where Vp' is the volumetric displacement of the pump per radian of pump rotation
per radian of swashplate angle measured from zero (swashplate vertical,
piston stroke zero)
is the swashplate angle measured from zero.
If allowing for variable displacement, the inertia of the swashplate and its
driving mechanism may have to be considered, whether or not an assumption of constant
pump rotational speed is assumed.
A pump invariably discharges into some kind of a volume. The volume will
include the pump's own porting and outlet, and can include the volumes of connecting
hydraulic lines, of a high-pressure filter, and an accumulator. The capacitance
effect of the pump's discharge side volumes can be recognized as a C effect, as
illustrated on Fig. 7.11d, where
Cp recognizes the capacitance effect of the discharge side immediately

adjacent to the pump
AQp is the amount of Qp 'lost' to the Cp effect
Qpx is the effective volumetric flowrate experienced by the next effect or
component to be added to the bond graph.
The majority of hydraulic systems include a pressure relief valve adjacent to the
pump, and this can be allowed for as an Rr effect as shown in Fig. 7.11e, where
Rr denotes the presence of the rel~ef valve

Qr is the discharge to tank through the relief valve,
both as discussed in Section 7.6, and
Qpy is the effective flowrate received by the next component or effect to

be added to the bond graph.
Either, both, or neither Cp and Rr may exist in a particular analysis.

124
R e m e m b e r i n g t h a t a d j a c e n t s i m i l a r j u n c t i o n s can be combined, leakage, discharge

side c a p a c i t a n c e , and r e l i e f v a l v e can be c o m b i n e d into Fig. 7.11f. N o t e that if
the Cp e f f e c t is c o n s i d e r e d n e g l i g i b l e , the causal b a r o n the o u t p u t b o n d Pp Qpy

m u s t r e v e r t to the left e n d o f the b o n d in o r d e r to p r o v i d e the single causal bar
n e e d e d a t the 0 - j u n c t i o n .
Fig. 7.11g shows a d e t a i l e d and g e n e r a l b o n d g r a p h for a p o s i t i v e displacement

pump, w i t h p o w e r in as T - ~p and o u t as Pp Qpy . The p u m p ' s e s s e n t i a l power-
t r a n s f o r m i n g c h a r a c t e r i s t i c is shown as the TF Vp(O, Pp) , where Vp(8, Pp) allows
that a p u m p ' s g e o m e t r i c d i s p l a c e m e n t rate can b e a f u n c t i o n o f 0 , a displacement
control p a r a m e t e r such as s w a s h p l a t e angle, and a l s o of the p u m p d e l i v e r y pressure
Pp i n the case o f p r e s s u r e c o m p e n s a t e d pumps.
I n c l u d i n g i n t e r n a l l e a k a g e t a p r e s s u r e - c o m p e n s a t e p u m p has a Qp - P p relation-
s h i p o f the form i d e a l i z e d on Fig. 7.12. A t c o n s t a n t p u m p speed, Q~p is insensitive
to pressure, in the m a n n e r o f a n o r m a l p o s i t i v e d i s p l a c e m e n t pump, u n t i l a set pressurl
Pset is reached. P r e s s u r e in e x c e s s of Pset is u s e d to activate a m e c h a n i s m which
c a u s e s p u m p d i s p l a c e m e n t and c o n s e q u e n t l y Qp to d e c r e a s e v e r y sharply. When Pp
falls below Pset , the p o s i t i v e d i s p l a c e m e n t p u m p c h a r a c t e r i s t i c is resumed. The
r e l a t i o n s h i p of Fig. 7.12 is r e a d i l y d e s c r i b e d as
QP = Vp ~p for Pp < Pset

7.18
= Vp ~ p + K p ( P p - Pset) for Pp > Pset
where Kp is the slope of the Pp > Pset p a r t o f Fig. 7.12. Note t h a t Kp is a

n e g a t i v e coefficient, of u n i t s (m3/s)/Pa .
M o s t o f the R , C , and I e f f e c t r e l a t i o n s h i p s i m p l i e d on Fig. 7.11 have

a l r e a d y b e e n d i s c u s s e d i n the p r e s e n t Chapter. O n l y p u m p friction r e q u i r e s further
discussion.
P u m p f r i c t i o n a r i s e s from
motion between mated mechanical parts within the pump, such as p i s t o n s in

t h e i r bores, c y l i n d e r b l o c k s on ~ o r t p l a t e s , p i s t o n - t o - s w a s h p l a t e bearings,
g e a r s r u n n i n g b e t w e e n side p l a t e s , v a n e s on t h e i r r u n n i n g tracks, etc.
shaft bearings
m o v e m e n t o f control m e c h a n i s m s i n the p u m p
fluid shear w i t h i n the p u m p due to m o v i n g parts.
P r e c i s e c o m p o n e n t - b y - c o m p o n e n t a s s e s s m e n t of friction i n p u m p s is a c o m p l i c a t e d
125
process. However, it can be appreciated that the torque required to overcome friction
is affected by both pump speed, and pump pressure, lhe moving parts will be i~nersed
in oil, providing good lubrication. Pump speed is likely to induce a viscous friction
effect, i.e. the higher the speed the higher the frictional torque, pump pressure
will increase the normal loading between mated mechanical parts in relative motion,
and thus increase the effort needed to overcome friction. That is, in general,
Tfp = f(mp, Pp) 7.19(a)
Normally, a pump's speed will not vary much, and thus the ~p component of Tfp will
be fairly constant. Pump pressure can vary widely and quickly as a system operates.
Thus, the pp component of Tfp can be dynamic. All things considered, if pump
friction needs to be included in a particular dynamic analysis, it can be represented
Tfp = Tfp~ + Kfp Pp 7.19
where Tfp is the total torque required to overcome friction

Tfp~ is a constant component of Tfp
Kfp is a coefficient
Pp is pump delivery pressure.
7.8.2 Coefficients
A specific pump's geometric parameters, such as Vp , Ip , and Kp are

obtainable from manufacturer's data. Leakage was described in Section 7.5, and
capacitance in Section 7.2. Internal friction coefficients Tfpm and Kfp are
more difficult to specify precisely, although an assessment of overall friction can
be made from mechanical efficiency data.
7.9 4-WAY CONTROL VALVES
7.9 .i Introduction
Because of
its position as the controlling element of many hydraulic control systems

its significance to the dynamic response of driven loads
its numerous design and performance variations
the 4-way control valve will be discussed in some detail.
The 4-way valve has four ports through which power can enter or leave the valve.
One port accepts power as an input from the pump system and is designated the supply
port. Another port is connected to exhaust (tank). The remaining two, the control
ports, typically but not exclusively are connected to the two hydraulic ports of a
hydraulic actuator. The direction in which a driven actuator moves is dictated by
126
the direction in which the control valve is actuated. The speed at which a driven
actuator moves is dictated b y the amount b y which the control valve is opened.
There are numerous functional variations of the 4-way control valve. It is

usual to characterize a valve by the port interconnections existing w h e n the valve is
in its central (neutral) operating state. The most common valve-centre states,
illustrated on Fig. 7.13, are
closed centre: all four ports are closed and are isolated from each other,
open-centre: all four ports are open to each other,
tandem-centre: supply port is open to tank (port), b u t the control ports

are both closed.
Various manufacturers have different ways of putting the illustrated porting

arrangements into physical hardware. There are centred valve port interconnections
which are different from those on Fig. 7.13. A 4-way valve can have two operational
positions (no centre position intended), can have three intended positions (valve
actuation full left, or central, or full right), or can be intended for continuous
actuation (caused to operate in any opening state over its full range of actuation).
A 4-way control valve can be of poppet type, or disc type, or spool type. It is not
practical to consider all of the variations and ramifications in the present Text.
A n appreciation of the control relationships for spool-type realizations of the basic
configurations of Fig. 7.13 will b e provided. From this, development of relationshi~
for variations in the basic valve functions should be practicable.
7.9.2 The Basic Relationship
Flow through a control valve is controlled b y the degree of opening of a set of

orifices formed b y the valve body-spool confi~iration. The rate of flow through an
orifice depends on the area of the orifice and on the pressure drop across it. This
functional relationship can be expressed
Q = f(Am, Ap n) 7.20(a)
where Q is flowrate through the orifice

A is the area of orifice opening normal to flow
AP is the pressure drop across the orifice
m and are indices.
In a well-made square-ported spool valve orifice of the type illustrated on

Fig. 7.14 it is a reasonable assumption that
A = K ' X
127
Y Z
a. Closed Centre
I
S
I
E
Y Z S Supply Port
IIII i
E Exhaust Port
b. Open Centre
Y" Control Port
I I Z Control Port
S E
Y Z
i I
C. Tandem Centre
S E
FIG. 7.13 CENTRE ARRANGEMENTS OF COMMON 4-WAY CONTROL VALVES
Circumferential Port Belt /

//'I // / I / /
Spool Lan
J-
FIG. 7.14 ILLUSTRATING SQUARE-PORTED SPOOL VALVE
For line-on-line, W = Wx
For overlap, Wx > W and overlap = 0.5(Wx-W)
For underlap, Wx < W and underlap = 0.5(W-Wx)
(Assuming axial symmetry)
128
m = 1
n = 0.5 (turbulent flow t h r o u g h an ideal orifice)
giying
Q = f(K X AP ~) 7.20(b)
where K is the spool l a n d c i r c u m f e r e n c e

X is the axial o p e n i n g caused b y the d i s p l a c e m e n t o f the spool in the valve
b o d y from the n e u t r a l (centre) position.
The f u n c t i o n a l e q u a t i o n 7.20(b) can u s u a l l y be e x p r e s s e d
Q = Kv X A~ % 7.20
where Kv is an overall c o e f f i c i e n t for a p a r t i c u l a r valve orifice, embracing port

c i r c u m f e r e n c e and d i s c h a r g e coefficients.
Valve coefficient Kv can v a r y s o m e w h a t w i t h degree o f o p e n i n g X , and with

pressure drop BP . However, i t is o f t e n adequate to assume Kv c o n s t a n t at a
m e d i a n v a l u e for the p a r t i c u l a r valve o r orifice.
Noting that it e x p r e s s e d flowrate Q as a f u n c t i o n o f p r e s s u r e d r o p AP ,

e q u a t i o n 7.20 c a n b e r e g a r d e d as a r e s i s t a n c e equation, and the valve o r i f i c e s as R
effects. E q u a t i o n 7.20 is n o n - l i n e a r in two ways; first because AP ~ is not of
degree 1 , and second b e c a u s e Q is a f u n c t i o n of the p r o d u c t of two variables, X
and AP . Thus, the "resistance" o f the o r i f i c e w i l l not b e a simple c o n s t a n t as in
c a s e s of l i n e a r resistance. In essence, the control v a l v e o r i f i c e s act as variable
n o n - l i n e a r r e s i s t o r s w h i c h can b e d e f i n e d
R = I / ( K v X " AP -%) 7.21
s u c h that
Q = I / R ~P 7.22
Note that R varies with valve opening X . The l a r g e r X , the l o w e r R becomes

a n d the l a r g e r the f l o w r a t e p a s s e d b y the orifice, and v i c e versa.
T h e i d e a t h a t e a c h flow o r i f i c e i n a 4 - w a y control valve acts as a r e s i s t o r

w h o s e r e s i s t a n c e is m o d u l a t e d b y v a l v e d i s p l a c e m e n t Xv w h i c h r e s u l t s in p o r t
opening X , is a useful w a y o f u n d e r s t a n d i n g the nature o f flowrate control imposed
b y such valves. It is f i r s t n e c e s s a r y t h a t all of the flow p a t h s of a particular
control v a l v e c o n f i g u r a t i o n b e recognized. Then the g o v e r n i n g flow e q u a t i o n s can be
formed. T h i s idea w i l l now be d e v e l o p e d i n turn for the three b a s i c configurations
of Fig. 7.13.
129
7.9.3 Valve Flow Nomenclature
The following nomenclature follows the general nomenclature of Chapter 2. It

is detailed here for convenience in appreciating the important characteristics of
control valve flows.
Qs flowrate into the supply port S from outside the valve

flowrate from the exhaust port E
QY flowrate from the control port Y
Qz flowrate from the control port Z
Qsy flowrate from supply into the Y port, controlled by orifice Rsy
Qsz flowrate from supply into the Z port, controlled by orifice Rsz
Qye flowrate from supply port Y to exhaust controlled by orifice Rye
Qze flowrate from supply port Z to exhaust controlled by orifice Rze
Qse flowrate from support port S to exhaust controlled by orifice Rse
Ps pressure in the supply port S
pe pressure in the exhaust port E
Py pressure in the control port Y
Pz pressure in the c ~ t r o l port Z
xv displacement of the spool from its centred position
x actual axial opening between spool land edge and port
x(o) amount of spool-to-port underlap, if present
W axial width of control port circumferential belt in valve body
Wx axial width of spool land (.'. overlap = 0.5 (Wx - W) )
Kv flow coefficient for control ports
Kse flow coefficient for supply-to-exhaust passage, if present
R denotes a resistive effect
It should be appreciated that except for coefficients X(o) , W , Wx , Kv ,

Kse , the quantities listed are dynamic, i.e. it is expected that they will vary
during dynamic response. Exhaust pressure Pe will normally be constant (i.e. a
source), and supply pressure Ps may in some circumstances be assumed constant.
7.9.4 The Closed-Centre Control Valve
Fig. 7.15a illustrates the closed-centre valve requirement. Fig. 7.15b

illustrates how the requirement can be realized in spool valve hardware. Ideally,
this line-on-line configuration allows no flow into or from the control ports while
the spool is centred. In practice, some overlap of spool lands over the port edges
is necessary to minimize leakage in neutral. This situation will be considered
after the line-on-line case.
Allowing that the spool can be moved either left or right of centre, there are
four flow controlling restrictions available in the valve, as illustrated on Fig.
7.15c. If the spool is central, each of the four possible flow paths have infinite
resistance, and no flow takes place. If the spool is moved left of centre, Xv
positive Fig. 7.15d, the resistance of openings Rsy and Rze becomes finite,
allowing flow, while the resistances of openings Rye , Rsz remain infinite,
preventing flow. Thus, flow can take place from supply through control port Y ,
and through control port Z to exhaust. Had the spool been moved right of centre
130
( Xv negative), Rsz and Rye would become finite, allowing flow from s u p p l y to
port Z and from port Y to e x h a u s t .
Fig. 7.15e illustrates the flow paths of the 4-way closed-centre valve in a
rectangular form which is u s e f u l for c o n s t r u c t i n g the valve's bond graph.
Fig. 7.15f shows a bond graph representation o f the c l o s e d - c e n t r e valve function.
It is a x i o m a t i c that the four control resistances are each a function of spool
displacement Xv . The basic equations implied on the bond graph are
Effect Equation
Rsy; Qsy = f(APsy): Qsy = Kv Xv APsy 7.23(a)
Rsz; Qsz = f(APsz): Qsz = Kv Xv APsz ~ 7.23(b)

%
Rye; Qye = f(APye) : Qye = Kv Xv APye 7.23(c)
Rze; Qze = f(APze): Qze = Kv - Xv APze % 7.23(d)
Supply 0-junction: Qs = Qsy + osz 7.23(e)
Exhaust 0-junction: Qe = Qye + Qze 7.23(f)
Y 0-junction: Qy = Qsy - Qye 7.23(g)
z o-junction: Qz = Qsz - Qze 7.23(h)
Rsy 1-junction: APsy = Ps - P y 7.23{i)
Rsz 1-junction: APsz = Ps - pz 7.23(j)
Rye 1-junction: APye = P y - Pe 7.23(k)
Rze 1-junction: APze = Pz - Pe 7.23(~)
The junction equations, (a) to (), are general. However, the ~ ~<~uations,
(a) to (d), are in p r a c t i c e discontinuous. Each needs to b e e x p a n d e d to e m b r a c e
physical reality, as f o l l o w s
Qsy = 0 if xv is 0
= Kv Xv APsy if Xv is positive and ~ w

7.23 (m)
= Kv W - APsy ~ if Xv is positive and > W
= 0 if Xv is negative
Qsz = 0 if Xv is 0
= 0 if Xv is positive
7.23(n)
= Kv IXvl APsz ~ if Xv is negative a n d ~ 97
= Kv W APsz ~ if Xv is negative and > W

131
a. Symbolic Configuration b. Physical Configuration

S
P//,,'/////
E Z
C, The Flow Paths, and Their d. Spool moved left ( Xv Positive).

Controlling Variable Rsy , Rze finite, permit flow.
Restrictions Rsz , Rye infinite, prevent flow.
S ~ ~ Y
~ Xv
E Z
,
Flow Qsy _ ~
through Rsy through Rze
e. Pressure-Flow- f. General Bond Graph
Resistance Circuit

QYel Apye ~,.Rye
Qye
PS
~sy P
s Be
Qs
'Rsz
1
Qs~l ~ R~
f Qz
Rze
pe~ Ps
E Qe " ~ze S Qs LlOPs
g, Bond Graph for Xv
Positive Only
Rsy h. Overlap E~L~ 0p e Qszl APszNR~z
~Psy~ I wx
pz01 Pz ~ Z
__--=--% Qze Qz
7 Qzel ~Pze ~:~e

APze~
T
Rze
FIG. 7.15 DEVELOPMENT OF BOND GRAPH FOR
CLOSED-CENTRE 4-WAY CONTROL VALVE
132
Q~e = 0 if xv is 0 |
= 0 if xv is positive
7.23(0)
= Kv IXvl APye ~ if Xv is negative and ~ W
= Kv W APye % if Xv is n e g a t i v e and > W
Qze = 0 if Xv is 0
= K v Xv APze % if Xv is positive and ~ w
7.23 (p)
= Kv W APze if xv is p o s i t i v e and > w
E q u a t i o n s 7.23(e) to (p) c o m p r i s e the model.
If s p o o l - t o - p o r t o v e r l a p is present, the b o n d graph is unaffected. However,

equations (m), (n), (o), and (p) m u s t b e a p p r o p r i a t e l y m o d i f i e d . For example, if W
is p o r t width, Wx is land width, a n d o v e r l a p is syranetrical (Fig. 7.15h), then
o v e r l a p is 0.5 (Wx - W) and the p o r t o p e n i n g b e c o m e s
X = 0 for IXv i < 0.5 (Wx - W)

X = X v - 0.5 (Wx - W) for Xv > 0.5 (Wx - W) and Xv positive
X = X v + 0.5 (Wx - W) for IXvl > 0.5 (Wx - W) and Xv negative
a~d these e x p r e s s i o n s m u s t be u s e d for Xv i n e q u a t i o n s 7.23 (m), (n), (o), and (p).
If control v a l v e a c t u a t i o n is restricted, as it o f t e n is for a p a r t i c u l a r sys~m

r e s p o n s e analysis, the g e n e r a l b o n d g r a p h o f Fig. 7.15f c a n b e g r e a t l y simplified.
For example, for Xv p o s i t i v e o n l y the simple b o n d g r a p h o f Fig. 7.15g is adequate.
It implies the necessary e q u a t i o n s
Qs = Qsy = f(APsy) = Kv Xv APsy -~|

Qe = Qze = f(APze) = K v Xv ~Pze ~
7.24
~Psy = Ps - Py
APze = Pz - Pe
This is a g o o d d e m o n s t r a t i o n of the advice "use the s i m p l e s t m o d e l w h i c h is
a d e q u a t e to the p a r t i c u l a r task".
7.9.5 The O p e n - C e n t r e Control V a l v e
Fig. 7 . 1 6 a i l l u s t r a t e s the c e n t r e d state of the o p e n - c e n t r e valve. Fig. 7.16b

i l l u s t r a t e s its h a r d w a r e r e a l i z a t i o n i n a spool valve configuration. Spool land
u n d e r l a p r e l a t i v e to the w i d t h o f the c o n t r o l p o r t s a l l o w s all four p o r t s to be open
to each o t h e r in the c e n t r e d - s p o o l state. The u n d e r l a p p e d spool lands a l l o w the
f o u r flowrate control r e s t r i c t i o n s Rsy , Rsz , Rye , and Rze , all to be finite
i n the c e n t r e state. Normally, t h o u g h n o t necessarily, all four R's will be equ~
133
in resistance w h e n the spool is central. N o r m a l l y also, in the central p o s i t i o n

Xvz0
Qs = Qsy + Qsz , and Qsy = Qsz = Qs/2

Qye = Qsy , and Qze = Qsz
Qe = Qye + Qze , Qe = Qs , and Qye = Qze
Qy = oz = 0
If the spool is m o v e d left, the Rsy and Rze o p e n i n g s will i n c r e a s e i n a r e a

and decrease i n r e s i s t a n c e to flow, to a l l o w i n c r e a s e d f l o w r a t e through themselves.
H~wever, Rsz and Rye w i l l increase in r e s i s t a n c e r e s t r i c t i n g flow t h r o u g h them-
selves. Then, Qy and Qz can b e c o m e finite, Qy p o s i t i v e and Qz negative, and
the valve can d r i v e an actuator.
If the spool is m o v e d further left, Rsz and Rye w i l l close, their r e s i s t a n c e s

becoming i n f i n i t e l y high, and flow through them b e i n g blocked. Meanwhile, Rsy and
~e can c o n t i n u e to o p e n u n t i l their p o r t s are fully o p e n to o f f e r m i n i m u m r e s i s t a n c e
flow. The v a l v e b e h a v e s like a c l o s e d - c e n t r e v a l v e a f t e r Rsz and Rye close.
If the spool is m o v e d r i g h t from centre, similar b u t r e v e r s e d a c t i o n s develop.
Note that Qsy d e s c r i b e s a flowrate through Rsy from supply into p o r t Y ,

but is not the flowrate from p o r t Y to outside, as p a r t of Qsy flows t h r o u g h Rye
become Qye . similarly, Qz and n o t Qsz is the external flowrate from p o r t
Z .
Fig. 7 . 1 6 c s h o w s a f u n c t i o n a l l a y o u t b e t w e e n the four p o r t s and the four flow-

controlling r e s t r i c t i o n s of the o p e n - c e n t r e valve. It is a v e r y u s e f u l bridge to
formation o f the b o n d graph of Fig. 7.16d. Figs. 7.16c and d are identical w i t h
their c o u n t e r p a r t s in Fig. 7.15 for the c l o s e d - c e n t r e valve.
General e q u a t i o n s for the o p e n - c e n t r e v a l v e are, as for the c l o s e d - c e n t r e valve,

discontinuous. For a s y m ~ e t r i c a l l y u n d e r l a p p e d v a l v e the m o d e l c o m p r i s e s the
junction equations, i d e n t i c a l w i t h those for the c l o s e d - c e n t r e v a l v e
Qs = Qsy + Qsz 7.25(a)

De = Qye + Qze 7.25(b)
Qy = Qsy - Qye 7.25(c)
Qz = Q s z - Qze 7.25(d)
APsy = Ps - P y 7.25(e)
~Psz = Ps - Pz 7.25(f)
~Pye = P y - Pe 7.25(g)
~Pze = Pz - Pe 7.25(h)
134
plus the R effect equations appropriate to open-centre with uniform underlap x(o)
and port width W
spool centred: Qsy = Kv'X (o) "APsy if Xv is 0

port part open,
= ~v- (x<o) x~) -APsy % if Xv is positive but < (W-X(o))
positive Xv:
%
port full open: = Kv'W "APsy if Xv is positive but ~ IW-X(o)) 7.25 (i)
port part open,
= Kv'(X(o)-IXvl)'APsy if Xv is negative but I<} X(o)
negative Xv:
port closed: = 0 if Xv i s negative but I>l X(o)
spool centred: Qsz = Kv.X(o).~Psz ~ if Xv is 0

port part open,
= Kv. (xco)-xv) . ~ p s z ~ if Xv is positive but < X(o)
positive Xv:
port closed: =0 if Xv is positive but X(o) .7.25(j)
port part open,
= Kv01X(o)+IXvl)oApsz if xv is negative but l<I IW-X(o)
negative Xv:
port full open: = Kv'W'APsz ~ if Xv is negative but I~) [W-X(o) 1
spool centred: Qye = Kv'X{o)'APye if Xv is 0

port part open,
= ~v. (xCo>-xv). ~Pye ~ if Xv is positive but < X(o)
positive Xv:
port closed: = 0 if Xv is positive but X(o) -7.25(k]
port part open,
= KV' [X(o)+IXv))'APye if Xv is negative but I<I IW-X (o) i
negative Xv:
port full open: = Kv.W.APye ~ if Xv is negative but 191
spool centred: Qze = Kv0X(o).APze if Xv is 0

port part open,
= Kv. (x~o) xv) -aPze ~ if Xv is positive but < (W-X(o)~
positive Xv:
port full open: = Kv'W. APze % if Xv is positive but ~ (W-X(o)~ 7.25 (}
port part open,
= KV" (X(o)-IXvl)'APze if Xv is negative b u t I<l X(O)
negative Xv:
port closed: = 0 if Xv is negative but I~I X(o)
Equations 7.25(a) to (j) provide a model allowing for all degrees Df valve
opening, positive and negative and with port opening saturation. A particular
analysis might not require comprehensive description of the valve characteristics,
and the bond graph and the model could be simplified accordingly, as indicated with
the closed-centre valve.
Particularly with open-centre valves, Kv may not be a single constant. They

often have a high gain adjacent to neutral and a lower gain as the spool moves well
clear of neutral. Gain here means the AQ - AXv characteristic for particular
values of AP .
135
a. Symbolic Configuration b. Physical Realization Sho%~

with Spool Centred
s
/ / /////
E Z
Qye through Rye-J Y / 1 Z 4-Qze through P~ze

! L .....
Qsy through Rsy -/ Qsz through Rsz
d. Bond Graph, drawn directly

f r o m c.
l
QYel Apye ~, Rye
c. Fl~;-Resistance Circuit
(same as for C l o s e d -
C e n t r e Valve)
PY 0 1 PY ~ y
QY
Qye PY
~e
1
Qy Y Q~Yl ap~v .:~y
Rs
~Qs,
s Ps
Qs J ": 0 P~
Pz Z
Qz
E~OPe ~szI ~Psz~,R~z
E Pe ~ -~ Rze~
]-Qz
Qe Qze
Pz ~ z
Pz : Qz
Qze
Qzel APze L: Rze
FIG. 7.16 DEVELOPING A BOND GRAPH FOR AN OPEN-CENTRE

4-WAY CONTROL VALVE
136
7.9.6 The T a n d e m - C e n t r e Control Valve
Fig. 7.17a i l l u s t r a t e s the t a n d e m - c e n t r e state, and Fig. 7.17b a realization in
spool v a l v e hardware. In the c e n t r a l p o s i t i o n the two control p o r t s Y and Z are
blocked and isolated while s u p p l y can f l o w fairly freely to exhaust.
The t a n d e m - c e n t r e valve has the four flow c o n t r o l l i n g orifices Rsy , Rsz ,
Rye , Rze , of the closed and open-centre valves, plus a fifth o r i f i c e Rse through
which flow can p a s s from supply to exhaust.
Fig. 7.17c illustrates the o r i f i c e s and f l o w p a t h s of a t a n d e m - c e n t r e valve,

in w h i c h
in the centre p o s i t i o n , only Rse is finite, the r e m a i n i n g four restrictions

having infinite resistance to flow. Thus only Qse is finite, and
Qse = Qs = Qe ,
as the spool is m o v e d left Rse remains finite though of i n c r e a s i n g

resistance to flow; Rsy and Rze become finite, permitting flows Qsy
and Qze ; Rsz and Rye remain of infinite resistance to flow,
as the spool is m o v e d further left, Rse becomes of i n f i n i t e resistance

preventing flow Qse ; Rsy and Rze remain finite, and Rsz and Rye
remain of infinite resistance,
the c o m p a r a b l e situation for spool d i s p l a c e m e n t to the r i g h t of centre can

be described.
Fig. 7.17d shows a b o n d g r a p h for the t a n d e m - c e n t r e valve. It is s i m i l a r to

those for the c l o s e d and o p e n - c e n t r e control v a l v e s except t h a t the s u p p l y to exhaust
path through the centre p o s i t i o n is i n c l u d e d as R effect Rse . The e q u a t i o n s
describing the flow p o s s i b i l i t i e s , prepared using the b o n d graph, are
0-junctions: Qs = Q s y + Qsz + Qse 7.26(a)

Qy = Q s y - Qye 7.26(b)
Qz = Qsz - Qze 7.26(c)
Qe = Qye + Qze + Qse 7.26(d)
1-junctions: APsy = Ps - Py 7.26(e)
APsz = Ps - Pz 7.26(f)
Apye = Py- Pe 7.26(g)

APze = Pz - Pe 7.26(h)
APse = Ps - Pe 7.26(i)
137
a. Symbolic Configuration b. Physical Configuration

s
xv
E
d. Bond Graph

Qye I APye~ ~
C. Flow-Resistor Circuit TQY e

Py 0: PY ~ Y
~ Rye
Osy I ~psy ~,Rsy
dRsy
S Ps~ s Ps 3 ~ N P~
Qs , Rsz
--~----~ Z Pe I~ ! T ~ A
~'Rze Qz
InPse T
E Pe Rse ~sz
Qe ~
Pz01 PZ ~ Z
Qz
Qze ~Qze
Qz~1 ~pz~ ~Rze

I
FIG. 7.17 DEVELOPING A BOND GRAPH FOR A TANDEM-CENTRE

4-WAY CONTROL VALVE
R effects: Qsy = 0 if Xv is 0 7
= Kv'Xv-APsy if xv is positive but < W
Kv.W.APsy ~ if Xv is positive but > W 7"26(9)
Qsz = 0 if Xv is 0
7 2 6 kl
= Kv-lXvl.APsz % if Xv is negative but < W
= Kv-W-~Psz ~ if XV is negative but ~ W
Qye = 0 if Xv is 0 I
= ~v Ixvl'~Ye if Xv is negative but < W 7.26()
= Kv W APye if Xv is negative but ~ W
Qze = 0 if Xv is 0
= Kv. Xv.APze ~ if Xv is positive but < W
7.26 (m)
= Kv.W.APze ~ if Xv is positive but 9 W
Qse = Kse Wse. APse ~ if Xv is 0

: ~se- (Wso-l~vl) . ~ s e ~ iflXvlis finite but < Wse 7.26 (n)
: 0 if IXvlis finite but ~ Wse
where line-on-line spool to port configurations have been assumed, Xv is spool

displacement from centre, w is port width, and Wse describes the opening of the
supply to exhaust port in neutral. The likely difference of the Rse port is
indicated by using specific Kse and Wse for it. Like the closed-centre valve,
overlap is likely in which case the X term and conditions for equations 7.26(j) to
(n) must be modified appropriately. If considered appropriate, the Kv values can
be different for each or some of the S , E , Y , and Z ports. Like the previous
control valves, the bond graph and its equations are simplified if valve motion is
restricted to one side only.
7.9.7 Summary
Bond graphs for closed-centre, open-centre, and tandem-centre 4-way control

valves have been presented. From them, equations appropriate to a particular
analysis can be written. The forms of the various equations to be expected have
b e e n given. Analysis of a specific valve under specific operating conditions must
include a careful selection of the bond graph to be used, and equally careful
specification of the equation set. The equations must recognize any valve lap
present, limits of port width opening, and any unsymmetrical aspects of the specific
139
valve. A restricted analysis often allows considerable simplification of a bond

graph and consequently of the equations needed.
Consideration of the bond graphs of Figs. 7.15, 7.16, and 7.17 reveals that
Fig. 7.17d could be regarded as a general bond graph for all three centre configurations
considered. For the closed and open-centre arrangements, it would be necessary only
to specify that the resistive effect Rse is infinitely high for all spool displacement
Xv , allowing no flow Qse . The equation set derived from the general bond graph
would be different for each of the three cases only in the Q = f(AP) equations.
Fig. 7.17d can be put into the circular form of Fig. 7.18a, or in the form of
Fig. 7.18b. qhis latter form might be the most handy, and will be used in future
examples.
Equations 7.20 to 7.26 show the types of equations and equation sets required to
describe the valve flows. Specification of valve opening X for each flow equation,
in terms of Xv (magnitude and sign), and of particular valve port and lap conditions,
is the major challenge. All flows, if finite, are unidirectional except for Qy and
Qz , the external flows at the control ports. If Qy is positive, Qz must be
negative, and vice versa, and the relevant equations must reflect this.
7 .i0 ACTUATORS
7. I0.1 Introduction
Bond graph structures for a linear actuator were developed in some detail in
Chapter 5, Section 5.3.2. Fig. 5.3 showed the developments for the conditions
back pressure effects neglected

no leakage across the actuator piston.
Inertia of the actuator's moving assembly (la) and its friction (Rfa) were
included in Fig. 5.3g, and neglected in Fig. 5.3h. In both cases the capacitance
effect of oil flowing into the cylinder is included. A linear actuator is included
in the system whose bond graph and equations were developed in Section 5.5.4, with
the same limitations associated with Fig. 5.3.
In the present Section, bond graphs for both linear and rotary actuators will be
further developed.
7.10.2 Linear Actuator
The back pressure developed in the discharge size of an actuator as its piston
is driven in one direction generates a force which opposes motion. This return-side
t40
a. A Circular Form
PY ~ y
Qy
Ps
Qs '~.0 ..... ":I I Q~e " O: Pe
Qe
~ E
Q~I ~ I/Qze
l Pz ~ Z
Qz
b. A Parallel Form
.................. PY ~ y
Rsy Rye
APsy~ Apy
~ 1 ~ --~ 0 ~,11 ~e'~

Ps Pe
as ":0 ~I' Qso ...........".......0 - - ~ e ~
IF---~O ~,II
APsz~ I APze~
Rsz Rze
............... Pz a,.- Z
Qz
FIG. 7.18 GENERAL BOND GRAPH, 4-WAY CONTROL VALVE

For closed and open centre Rse offers infinite
resistance, rendering QSE = 0 for all Xv .
Functionally identical with Fig. 7.17d.
141
effect can be included in a bond graph via the 1-junction shown on Fig. 7.19a, which
sums algebraically the forward-side pressure-generated force Fay and the return-side
pressure force Faz . The 1-junction merely describes the net pressure force Fa as
Fa = Fay - Faz 7.27
Note that the back pressure effect causes a power loss through a loss of effective
force.
Fig. 7.19b shows the development of Fig. 5.3g to allow for the back pressure
effect. Pressure Paz is generated because flow is being forced from the discharge
side of the actuator into the return line by forward motion of the actuator piston.
This is clearly a capacitive effect (flow into a C causes effort build-up in the C )
and this is recognized by including Caz in the return line bond graph. It is also
the reason why back pressure Paz is causal at the TF end of its bond.
If it was decided that actuator friction (Rfa) and inertia (Ia) could be
neglected in a particular analysis, or combined with the friction and inertia of a
driven load, they and their 1-junctions are simply deleted from Fig. 7.19b.
To allow for the effects of internal leakage across the actuator piston, the bond
graph can be modified to that illustrated on Fig. 7.19c. A bond from the 0-junction
of Cay , or from a separate 0-junction downstream of it, allows that some of the
incoming flow Qay is lost as leakage Qa . A bond into the 0-junction of Caz ,
or into a separate O-junction upstream of it, allows Qa to he added to Qaz . The
l-junction placed in the Qa flow bonds recognizes that it is pressure drop
(Pay - Paz) across the piston that induces leakage.
Fig. 7.19c is a quite detailed model structure for a linear actuator, and is
suitable for study of the dynamic performance of the actuator itself. It is suitable
als0 for connection to the bond graphs of inertia loads having compliance between
actuator and inertia such as were discussed with Fig. 7.5, Entries 4 to 7. Entry
and/or exit pressure loss can be included by adding 1-junctions at the left.
It is important to be aware that Fig. 7.19c and its causalities are not unique.
The actuator bond graph can depend also on the type of load it is to drive. For
example if it is to drive a simple inertia without compliance in the connection, the
causality of the force output bond on Fig. 7.19c will be incorrect. The inertia load,
Fig. 7.5 Entry i, requires the causal bar at the right end of the Fa bond of
Fig. 7.19c. A little thought resolves this conflict. If there is no compliance
between Ia and Im they should be lumped, reflecting the rigidity of connection.
The Ia bond of Fig. 7.19c becomes the actuator's output bond and the former Fa bond
is deleted as irrelevant, Fig. 7.19d. The same arrangement applies for the same
142
reason to the load systems represented on Fig. 7.5, Entries 2, 3, and 8. Actuator
friction and load friction also can be lumped together if there is no compliance
between them, i.e. if their bonds share the same velocity.
Note that the effects Cay and Caz cannot be represented by simple coefficient,
except for small perturbation response analysis. In general the cylinder end volumes
vary with their displacement as discussed in Section 7.2, and the formulation of the
Ca effects should reflect this.
In Chapter 8 there are further examples of various actuator-load situations.
7.10.3 Rotary Actuator (Hydraulic Motor)
The hydraulic motor can be viewed as the rotational equivalent of the linear
actuator. Its moving parts have inertia and friction, its chambers capacitance,
and it has internal leakage. Fig. 7.19 can be quickly modified to its rotational
equivalent by replacing force by torque, velocity by angular velocity, and the TF
modulator coefficients A by Va , where Va is the swept volume per radian of
rotation of the motor. The TF equations become
Pay Ha = T 7.28
Qay Va -I = ~ 7~29
7.10.4 Equations and Coefficients
The equation forms associated with Pig. 7.19 and its rotational equivalent have
all been discussed and will not be re-presented here. Attention is drawn to the
variation of the end volumes as the piston of a linear actuator moves. This affects
the capacitances of the linear actuator, as discussed in Section 7.2. The varying
capacitance effect does not occur to anywhere near the same extent with hydraulic
motors and can usually be ignored.
There is usually little difficulty in arriving at the values of coefficients for

actuator equations. Most are apparent from manufacturers' data sheets. Leakage
coefficients can be inferred from volumetric efficiency data, friction from mechanical
efficiency, while geometric data such as piston areas for cylinders, swept volume per
radian for motors, and inertias of moving assemblies are supplied directly. This
leaves estimation of capacitances Cay and Caz as the main difficulty. This in
turn requires estimations of volumes under pressure and effective bulk modulus, both
at the specific location being considered, as discussed in Section 7.2.
Loaded hydraulic actuators usually have low damping, with damping ratios of the
order of 0.01 to 0.2 depending on the conditions of bearings, seals, and glands.
143
a. Allowing for Back b. Including Actuator Capacitances (Cay and

Pressure Effect Caz), Friction (Rfa) and Inertia (Ia)
Cay
pqy Fa, Xa
ay ~! / pa;I mv' Supply I Pay.:
Fa
Paz ~ TF --~- Fay
Qay Ay Xa~
"1 ---~I---~I ~ ~a
Fa Return ~ ~J
1 ~a ~-- 0 Qa~Az
. / Ia
A Paz TF ~ a z
Qaz Az Caz
c. Including Internal Cross-Piston Leakage (Qa through Ra)

Cay
AQay
Supply ~ 0 PaY ~I TF Rfa

Qs Qay '
Qa Ay
Xa Fay Ffa
I ~Pa ~ R E a " ! Xa ~ Output
~Qa
Pe
"~AQaz Az Ia
Caz
d. Output Fnd of Bond Graph if Actuator Driving a Rigidly Connected

Inertia Load Im
(Rfa + Rfm)
------ ~ TF.
I ": (~m + ~
Az
FIG. 7.19 BOND GRAPHS FOR LINEAR ACTUATOR

144
7.11 SOME O T H E R C O M M O N C O M P O N E N T S (Lines, Filters, A c c u m u l a t o r s , Valves)
7.11.1 Hydraulic Lines
In m o s t h y d r a u l i c control s y s t e m s the l i n e s are s h o r t and are of r e a s o n a b l y

large d i a m e t e r to k e e p f l u i d flow v e l o c i t i e s to a low value. In these s i t u a t i o n s it
is u s u a l l y r e a s o n a b l e to ignore line r e s i s t a n c e as n e g l i g i b l e r e l a t i v e to, for example,
control v a l v e r e s i s t a n c e , line c a p a c i t a n c e as n e g l i g i b l e r e l a t i v e to, for example,
a c t u a t o r capacitance, and fluid i n e r t i a n c e as a v e r y m i n o r d y n a m i c e f f e c t r e l a t i v e to
the i n e r t i a of the a c t u a t o r and load. Alternatively, line r e s i s t a n c e can be lumped
w i t h v a l v e r e s i s t a n c e and p o s s i b l y w i t h a c t u a t o r i n l e t r e s i s t a n c e , and line capacitance
w i t h a c t u a t o r or o t h e r a d j a c e n t camponents' capacitance(s). There m a y be cases
h o w e v e r w h e r e it is d e s i r a b l e to include some or all of a line's R , C , and I
e f f e c t s in a d y n a m i c model. The s i t u a t i o n could arise
if the l i n e s are long

if the l i n e s are of small b o r e
if the system is b e i n g d r i v e n in a h i g h l y d y n a m i c mode.
P r e s s u r e d r o p in a line o f t e n can b e a c c o u n t e d for simply b y p l a c i n g a 1-junction

in the a p p r o p r i a t e p o w e r bond. For example, Fig. 7.20a shows line r e s i s t a n c e as Rh
i n c l u d e d in b e t w e e n a control v a l v e (Rv) and an a c t u a t o r (Ca) In this particular
situation, the Rv and Rh e f f e c t s come from a d j a c e n t 1 - j u n c t i o n s and can b e shown
o f f the one 1-junction. Rv and Rh r e a l l y s h o u l d n o t b e lumped as they are of
d i f f e r e n t causality. For Rv we want Qv = f(Rv) to give the c o n t r o l l e d flowrate
through the valve. For Rh we want APh = f(Qv) to a l l o w for p r e s s u r e d r o p in the
line. The Rh e f f e c t is u s u a l l y small and if n e g l e c t e d or lumped w i t h Rv no great
e r r o r i n s y s t e m r e s p o n s e results.
Similarly, line c a p a c i t a n c e can be r e p r e s e n t e d b y ch l e a d i n g from an 0-junction

i n the a p p r o p r i a t e p o w e r bond, Fig. 7.20b. As w i t h Rh , just w h e r e Ch fits in a
b o n d g r a p h d e p e n d s on the a d j a c e n t components. C e f f e c t s from a d j a c e n t 0 - j u n c t i o n s
(no 1 - j u n c t i o n s between) can be lumped. In fact, to p r e s e r v e integral c a u s a l i t y they
m u s t b e inmped (Fig. 7.20c) , r e f l e c t i n g that o n l y one v a l u e o f the p o t e n t i a l v a r i a b l e
can e x i s t around an 0 - j u n c t i o n C effect. It is p r o p e r to separate line capacitance
from a n o t h e r c a p a c i t a n c e o n l y if there is an R or I e f f e c t in b e t w e e n . For
e x a m p l e if there is a s i g n i f i c a n t a c t u a t o r e n t r y p r e s s u r e drop, Fig. 7 . 2 0 d could be
appropriate. It is of i n t e r e s t to see that e n t r y r e s i s t a n c e Ra has the same
c a u s a l i t y as R v , r e q u i r i n g d e s c r i p t i o n as Qa = f(APa) r a t h e r than the ~Ph = f(Qv)
o f the line r e s i s t a n c e Rh . In e s s e n c e , Ch b e c o m e s the d r i v i n g c o m p o n e n t as far
as flow i n t o Ca is concerned, and Ra the c o n t r o l l i n g resistance.
In e x t r e m e cases (long small b o r e lines, h i g h l y d y n a m i c flows) the line dynamics

can b e i n c l u d e d b y s e m i - d i s t r i b u t e d p a r a m e t e r m o d e l l i n g in w h i c h the R and C
145
a. Including Hydraulic Line R e s i s t a n c e (Rh)

Rv Rh Ca
T!
!
~I:
C. The I n c o m p a t i b i l i t y of S h o w i n g
Adjacent C Effects. (Where
b. Including Hydraulic Line can the c a u s a l b a r go on the
Capacitance (Ch) c e n t r e bond?)
Cl c2
~Qh
Ch ', -0 "0 ":

d. R e s i s t a n c e (Rh), C a p a c i t a n c e (Ch), E n t r y R e s i s t a n c e (Ra) in a
H y d r a u l i c Line B e t w e e n a C o n t r o l V a l v e (Rv) a n d A c t u a t o r (Ca)
Rv Ch Ca
"0 ~, b a d
Rh Ra
FIG. 7.20 BOND GRAPHS FOR HYDRAULIC LINES
Rb Cb
Pl P2 P2
FIG. 7.21 A L L O W I N G F O R THE R E S I S T I V E (Rb)

AND CAPACITIVE (Cb) E F F E C T S OF
A FILTER
b. RC Representation of an
a. Simple Capacitive
Accumulator
Representation of
C
an A c c u m u l a t o r
P2 = f(~Q)
AQ = f(AP)
~P = Pl - P2
Q2 = Q1 - AQ
~Q p = f (AQ) ~QI Ap or Q1 = Q2 + AQ
AQ = Q1 - Q2
P1 and either Q1 or Q2
P equations will arise from
adjacent component bonds.
FIG. 7.22 BOND GRAPHS FOR ACCUMULATORS

146
effects of the line are represented as a series of RC effects. Fluid inertiance

can be included in both the lumped parameter and the distributed parameter approaches.
Multi-lump distributed parameter models allow for the time delay effect associated
w i t h long lines.
Accurate dynamic modelling of long fluid lines is a complex subject beyond the
scope of the present Text. Ref. 7.7 provides a useful and relevant review.
7.11.2 Filters
A filter in a hydraulic line causes resistance (Rb) to flow which is reflected

as a pressure drop, and a capacitive effect (Cb) due to its fixed volume. This
effect is readily allowed for as the bond graph of Fig. 7.21 which implies the
relationships
Apb = f(Ql)
Pl = P2 + APb
P2 = f (AQb)
AQb = Q1 - Q2 7.30
Q1 = (to be determined from upstream component)

Q2 = (to be determined from downstream component)
Whether or not the Rb , Cb effect is significant enough to be allowed for in a

particular system analysis m u s t be decided by the analyst. W h e t h e r or not Rb can
be lumped with an adjacent R , possibly of the hydraulic line, depends on the
circumstances of the particular system and its bond graph. So too the Cb effect
m a y be able to be lumped with an adjacent downstream C effect such as that of an
actuator.
7.11.3 Accumulators
Accumulators are used in hydraulic control systems to provide some combination

of the functions
to reduce (damp out) flow and pressure pulsations in particular lines,
to provide supply flowrate, and hence power additional to the pumps, for
peak load demands o f short duration,
to maintain pressure in systems required to be inactive, b u t under pressure

ready for use, for long periods. In these systems, the pump charges the
accumulator as required and the accumulator supplies the system. The
p u m p can be operated under automatic cut-in and cut-out control from
pressure limit switches, or can be pressure-compensated if it needs,
147
to run continuously,
to soften the pressure transients in systems having rapid stop-start or

flow direction reversals.
Many hydraulic control systems do not have accumulators.
If an accumulator is simply providing additional volume to that of its associated

line, it can be represented as the C effect illustrated on Fig. 7.22a. Note that
the accumulator requires the flow-causal input bond (i.e. pressure is effect) and the
pressure-causal output bond illustrated on Fig. 7.22a. Insertion of an accumulator
in an existing system bond graph requires reassessment of causalities in the bond
graph.
If, either in effect or by design, there is a significant resistance to flow

into or from the accumulator, it behaves like the RC effect illustrated on Fig.
7.22b. The causal bars on the input (Pl QI) and output (Pl Q2) bonds depend
on the adjacent system components and could be the opposite to those shown.
The relevant functional relationships required have b e e n added to Fig. 7.22.

The C and R effects in no w a y imply linear relationships. However, unless there
is reason to think them inappropriate, the expressions
P = I/C - IQ dt + p(o) 7.31
where C = V/B
and AQ = I/R AP 7.32
could be used. It should be borne in m i n d that V , the volume of oil under pressure
in and around the accumulator, will vary depending on the state of charge of the
accumulator. It should be borne in m i n d also that for most system analyses, the
primary requirement will be to study the dynamics of the driven load, and not those
of the accumulator. Usually a reasonable approximation of the accumulator dynamics
is adequate. If a more accurate model of the accumulator dynamics is required, V
may have to be appropriately modulated, and R may have to be described by a square
root of pressure drop relationship.
Use of a large accumulator w i t h the supply can justify an assumption of constant

pressure supply to the control valve of a valve-controlled actuator system.
Accumulators ~ d their c h ~ a c t e r i s t i c s are discussed in detail in Refs. 7.8 and

7.9.
148
7.11.4 Some Other Valves
Pressure relief valves and directional control valves have been discussed in
Sections 7.6 and 7.9 respectively. It is not practicable to present here the bond
graphs of the large number of valve variations available commercially. Several will
be presented now, and it is left to the reader to form bond graphs for other valves
when they are required.
Check Valve
The check valve, or non-return valve, is intended to allow free flow in one
direction and to prevent totally flow in the other direction. Most check valves
Will have a flow-pressure drop characteristic of the kind idealized on Fig. 7.23a.
Negative pressure drop results in no flow. A small positive pressure drop is
usually present when flow is positive and this pressure drop can increase somewhat
with flowrate. Pressure drop in the presence of flow means that there is power
loss. Just how this is best accounted for on a bond graph depends on the location
of the check valve relative to its adjacent components.
For example, consider the system of Fig. 7.23b. There will be three pressures
involved when the load is being driven, these being supply Ps , pressure Px
between control valve and check valve, and pressure Pa downstream of the check
valve. To accommodate this the bond graph of Fig. 7.23o could be used. It
includes capacitance Cx between control valve and check valve. The causality of
the check valve Rx effect allows that flowrate through the check valve depends on
the direction and size of the pressure drop ~Px across it. Some appropriate
equations are included with Fig. 7.23c.
Fig. 7.23d shows another interpretation in which the pressure characteristic of

the check valve is taken as a source of pressure i.e. a pressure-effect input
the system which is substantially independent of the system. SPx would simply
describe the characteristic of Fig. 7.23a. It is not always necessary to include
check valves in limited dynamic analyses.
Counterba lance Valve
A counterbalance valve, Fig. 7.24a, is used to prevent an open-centre valve-

controlled actuator with a suspended vertical load from running away when the valve
is in neutral. Fig. 7.24b shows a typical application. The counterbalance valve
allows free flow from the control valve to the underside of the actuator for the load-
lifting stroke. It prevents flow from the actuator when the control valve is in
neutral. In effect the counterbalance valve will not open to allow discharge fl0w
until a certain set pressure is exceeded in the lower actuator line. This pressure
is enough to generate a force on the underside of the actuator piston large enough to
support the suspended load. If the control valve is moved to drive the actuator
149
a. Pressure-Flow Characteristic of Check Valve

+
APx
b. System with Check Valve
QI, Px Pa
c. Bond Graph Allowing for Power Loss at the Check Valve

Rv Cx Rx Ca
Ps
~1: '~ 0 ~:I: ~ 0 Q---'-~ TF
Ql Px Q2 Pa
Q1 = f(APv, Xv) APv = Ps - Px
Px = f (Qcx) Qcx = Q1 - Q2
Q2 = f (APx) APx = Px - Pa
Pa = f(Qca) Qca = Q2-Qa
d. Bond Graph when Check Valve Pressure Loss Taken as a Source
Rv RX Ca Rv~ _Rx Ca
i iQca Ps
v SP/ Qca
Zz---~l ~----"'- 1 : - 0 --'~'a-" ' or ~'~ 1 : ~ 0 . . .Qa
.
Q1 Q1 Pa Q1 pa
Q1 = f(APv, XV) APv = Ps - SPx - Pa

Spx = specified from check valve characteristic
f
Pa = f(Qca) , probably 1/Ca IQca dt + Pa(o)
etc.
FIG. 7.23 INCLUDING CHECK VALVE ON BOND GRAPH

150
a. Counterbalance Valve Function b. An Application

control Counterbalance Loade~
Valve Valve Ac%ma%~t
t \,, \/ '" ~,
C. Bond Graph, For Load Being Lifted
~-------"
PY
Qy
"*'
I:
"'-- - ,TF ---~ k
SFg
Ay %, 4
\ k
k~& \ I
Cay \ .~
Y- Fa
~1~a-~ I
.... ~ Im
4 caz ~a
T
SPx I / ,, // I
li " 0 -~a-"'-ITF ---J Rz

Paz Az
\ / \ %% ....
Check Vaive of Actuator Load
"~balance Valve
",,Control Valve
See Fig. 7.58
d. Bond Graph, For Load Being Driven Downwards
k Ay ~k i
k~Cay k%
wl ~,Caz i )~ Xa 3- Xa
SP // / I
/
'I" Qz
~= 1 ," !-)~
paz . Qaz
. . . ~TF "- --""
Az
Rf
\ /
Pressure-actuated Valve Qaz = 0 if Paz < SPw
of Counterbalance Valve = Xa - Az if Paz > SPw
Pz = Paz - SPw
SPw = source pressure
FIG. 7.24 INCLUDING A COUNTERBALANCE VALVE

151
down, the pressure build-up in the upper actuator line causes increased pressure in
the lower side and the counterbalance valve opens to allow discharge flow, and hence
downward motion of the load.
Fig. 7.24c shows a bond graph for the counterbalance valve for the condition
~at the load is being lifted. The check valve structure of Fig. 7.23d has been
adopted. That is, the check valve is seen as providing a particular form of
pressure source SPx . Supply Pz , Qz can be regarded as having come from the
Z port of control valve structure of the form of Fig. 7.18, while discharge Py ,
Qy is feeding into the control valve's Y port. The various equations appropriate
to Fig. 7.24c have all been developed previously.
Fig. 7.24d shows a bond graph for the counterbalance valve for the condition
that the load is being driven downwards. The pressure-actuated valve is set at a
certain threshold pressure. Thus, it can be shown on a bond graph with a symbol and
causality similar to that used for a check valve. In this case the R effect is
designated as the pressure source SPw . SPw will have a particular form which is
substantially independent of the system in which the valve is used.
It is probably possible to combine Figs. 7.24c and d into a single bond graph.
B0wever, the algebra of the equations would become confusing. It is probably better
to use the separate diagrams to prepare the joint equation set using the notions that
Fig. 7.24c applies when control Xv is negative, and Fig. 7.24d when Xv is
positive.
Counterbalance valves are sometimes pilot-operated by the pressure on the other

side of the actuator. If this situation exists, the equations have to be
appropriately modified.
7.12 CONCLUS ION
The more common components, actions, and phenomena relating to the dynamic
response of hydraulic control systems have been considered. The relationships
adopted or proposed are not necessarily unique. A particular analyst may prefer
different equation forms. Particular systems may require the use of additional
equations or of different equations to describe adequately what the analyst feels
will occur. With some experience in modelling, simulation, and in observation of
hardware system performance, the designer-analyst can develop modified bond graphs
and relationships. The objective of the present Chapter - - indeed of the present
Text - - is to provide an approach to modelling, enough bond graph material, and
enough equation forms, for simulation as a system design aid to be viable.
152
Quite apart from the deciding of equation forms, the selection of numerical
values for coefficients in the equations is not always easy. Component manufacturers
may have to be spec~fically solicited for some data.
7.13 REFERENCES
7.1 Merritt, H.E., Hydraulic Control Systems, Wiley, 1967.
7.2 Davis, D. and Stecki, J.S., Effective Bulk Modulus in Hydraulic Systems,
Proc. Fluid Power Conf., FPS, Sydney, 1978.
7.3 Dransfield, P. and Bruce, D.M., Leakage Flow Rata Past Pistons of Oil
Hydraulic System Components, J. of Aircraft, AIAA, VS, N2, 1968.
7.4 Barnard, B. and Dransfield P., Predicting Response of a Proposed Hydraulic

Control System, Using Bond Graphs, Trans. ASME, JDSMC, V99, Series G, Nl,
1977.
7.5 Chong, F.K. and Dransfield, P., The Effect of Choice of Relief Valve on
the Response of a Hydraulic Control System, Proc. ist Aust. Conf. on
Control Eng., I.E. Aust., 1979.
7.6 Steber, G.R. and Romer, I.C., Simulation of a Pressure Relief Valve,
Simulation, V12, N5, 1969.
7.7 Davis, D.C. and Stecki, J.S., Hydraulic System Analysis - - M o d e l l i n g of

Fluid Transmission Lines, BFPR Jrl., Vl4, N3, Oklahoma State University,
1980.
7.8 Knight, G.C., et al, Connection Capacitance Effects in Hydrostatic

Transmission Systems and their Prediction by Mathematical Model, Proc.
I.Mech.E., V186, 1972.
7.9 Svoboda, J., Analogue and Digital Modelling in the Design of a Hydraulic
Vehicular Transmission, Proc. I.Mech.E., V193, 1979.
CHAPTER 8
APPLICATIONS OF BOND GRAPHS
8.1 INTRODUCTION
Chapters 4 and 5 introduced the ideas of power flow and power bond graph
modelling respectively. Chapter 6 outlined the solution procedure once a model is
obtained. Chapter 7 presented and discussed bond graph structures for components
commonly used in hydraulic control systems. Application of the techniques will now
be demonstrated more adequately using a series of case studies escalating in complete-
ness and complexity. Where judged necessary, explanatory detail is provided.
8.2 CLOSED-CENTRE VALVE CONTROLLED INERTIA LOAD WI%~H FRICTION
Fig. 8.1a illustrates the system, which was used previously as an example in
Chapter 5 (Section 5.5.4, Fig. 5.12) . At that stage return line effects were
neglected, the model was restricted to positive spool displacement only, and it was
assumed that fluid was supplied to the valve at constant pressure. The example will
now be extended to include return line effects and to allow spool displacement in
both directions and thus load motion in either direction. The assumption of constant
pressure supply will be retained, though its validity is questionable.
Fig. 8.1b shows the components considered to be relevant to the present analyis.
Fig. 8.1c shows the bond graph, formed by direct coupling of the bond graphs of the
major components, taken from Figs. 7.18 for the valve (with Rse neglected), 7.19b
for the actuator (with Rfa and la deleted as they will be lumped with the load,
the actuator and load being rigidly coupled; and internal leakage Qa neglected) ,
and 7.5(2) for the load. Rsy , Rsz , Rye , and Rze have each been shown as Xv
a. The System
Xm
[ ~ J ~1~ ll;:lJ I,.,.. I ii~II

iiiiii
I//2/I
b. The Components to be Modelled Xm
S xv i
//III~I
C. ~e Bond Graph PY
Rf
/ ~xv) Rye(xv) ~
sPs
Qs ~ 0 sPs sPe0:~ ~ ~m I
Xm
~z(xv) ~e(xv) / F/
Pz
I \ I
4-Way Control Valve Linear Actuator Driven I~ad
FIG. 8.1 SYSTEM OF EXAMPLE 8.1

155
dependent, and Cy and Cz e a c h as Xm dependent. The Xm d e p e n d e n c e of Cy

and Cz could have b e e n shown on the b o n d graph as b r o k e n lines w i t h full a r r o w h e a d
leading from Xm to Cy and Cz , w i t h i n t e g r a t i o n o f ~ to show Xm .
R e s i s t a n c e o f the h y d r a u l i c lines has b e e n n e g l e c t e d or l u m p e d w i t h the valve

resistance, and line c a p a c i t a n c e has b e e n n e g l e c t e d or lumped w i t h a c t u a t o r capacit-
ance. The c o n d i t i o n of flow into the Y e n d and from the Z e n d o f the a c t u a t o r
has b e e n taken as p o s i t i v e i n assigning p o w e r flow d i r e c t i o n s . E a c h p o w e r state
variable is s h o w n once o n l y on the b o n d graph. Using the notions that there can be
only one e f f o r t v a r i a b l e on b o n d s around a p a r t i c u l a r 0-junction, and o n l y one flow
variable on b o n d s around a p a r t i c u l a r 1-junction, the r e a d e r c a n "see" the "missing"
variables. A s s i g n m e n t of c a u s a l i t y follows the p a t t e r n s used p r e v i o u s l y i n the Text.
H a v i n g d e c i d e d that the b o n d g r a p h structure of Fig. 8.1c a d e q u a t e l y reflects

the d y n a m i c effects r e q u i r e d in the model, i t is r e q u i r e d next to choose an e q u a t i o n
for e a c h p o w e r state variable. The equations, d e v e l o p e d in the o r d e r Sources, I
effects, C effects, R effects, Transformations, Junctions, are g i v e n i n Table 8.1.
The first 26 e q u a t i o n s are the p o w e r state equations. B o n d g r a p h flow directions,

and the equations, w e r e p r e p a r e d on the b a s i s that Xv w o u l d be positive. The
equations h o l d e q u a l l y for Xv negative. In that case Qy , Qya , F y , and Qz ,
Qza , Fz e a c h b e c o m e negative relative to their d i r e c t i o n s a s s i g n e d o n the b o n d
graph.
The m o d e l is c o m p l e t e d b y adding the four a u x i l i a r y e q u a t i o n s 8.1.27 to 30. To

solve (simulate) the model, values are r e q u i r e d for initial v a l u e s Xm(o) , Py(o) ,
Pz(o) , Xm(o) , and for coefficients SPs , SPe , Im , Kv , F~ , Kf , Ay ,
Az , Vy(o) , Vz(o) , B and the i n p u t Xv(t) n e e d s to be specified. The
simulation w o u l d p r o v i d e system response to the s p e c i f i e d input. It is likely that
the a n a l y s t w o u l d arrange for p l o t s of Xm(t) , Py(t) , Pz(t) , and p o s s i b l e
Qsy(t) and Qsz(t) , although o t h e r o r more v a r i a b l e s can b e s t u d i e d if required.
If the control valve had overlap, or i f it was o f o p e n - c e n t r e or t a n d e m - c e n t r e

configuration, the e q u a t i o n s are r e a d i l y adjusted u s i n g the d e t a i l s g i v e n i n
Section 7.9. The valve flow e q u a t i o n s 8.1.6, 7, 8, and 9 can be r e a d i l y e x p a n d e d to
allow for s a t u r a t i o n of p o r t opening, also as d e m o n s t r a t e d i n S e c t i o n 7.9.
8.3 LOADED HYDRAULIC SERVOSYSTEM
Fig. 8.2a i l l u s t r a t e s a type of s e r v o s y s t e m i n w i d e use in p r o p o r t i o n a l p o s i t i o n -

ing systems (machine tool slides, a i r c r a f t control surfaces, etc.). Asstuning
156
constant pressure supplyt an a s s u m p t i o n s t r e n g t h e n e d b y t h e p r e s e n c e o f t h e accumulator

u p s t r e a m of the c o n t r o l valve, the c o m p o n e n t s e s s e n t i a l to the d y n a m i c analysis are
s h o w n o n Fig. 8.2b. This c o n f i g u r a t i o n w a s m o d e l l e d c o n v e n t i o n a l l y i n Chapter 3
w i t h o u t the p r e s e n c e o f the e x t e r n a l force Fe now i n c l u d e d as p a r t of the load.
Fig. 8 . 2 o shows the b o n d g r a p h p r o p o s e d for the system. It is similar to

Fig. 8 . 1 c e x c e p t for
the allowance for c o m p l i a n t r e s t r a i n t of the load (force Fk on the Ck

bond}
allowance for the e x t e r n a l force Fe ( SFe bond)
combining of the a d j a c e n t Fa and Fm 1-junctions
i n d i c a t i o n that the v a l v e resistances are b o t h Xi (input) and Xm (load

displacement) dependent. Spool d i s p l a c e m e n t Xv will be b o t h Xi and
Xm dependent
Cy and Cz b e i n g s h o w n as Xm d e p e n d e n t v i a d i r e c t feedback of Xm .
The set of e q u a t i o n s is similar to the 8.1 set w i t h the m o d i f i c a t i o n s
the a c t u a t o r - t o - l o a d 1-junction{s) e q u a t i o n s 8.1.25 and 8.1.26 b e c o m e
Fm = Fy - Fz - F f - Fk - SFe 8.2.1
the Xv e q u a t i o n 8.1.30 b e c o m e s
xv = (L2 - LI)/L2 - Xi - L I / L 2 Xm 8.2.2
the i n p u t e q u a t i o n b e c o m e s
Xi = to be s p e c i f i e d 8.2.3
additional l o a d force e q u a t i o n s are
SFe = constant 8.2.4
Fk = i/Ck - fXm d t + Fk(o) 8.2.5

J
8.4 PUMP SUB-SYSTEM
M o s t h y d r a u l i c control systems require a p u m p i n g u n i t (power supply) consisting

of
the p r i m a r y drive (electric motor, diesel engine, a i r c r a f t engine, etc.)

a positive displacement pump (fixed or v a r i a b l e delivery)
157
TABLE 8.1 Model for Example 8.1
Component~ Effect, Comment Equation

power State Equations :
Supply, source SPs cons tant 8.1.1
Exhaust, source SPe constant 8.1.2
Im effect Xm i/Im - iFm dt + Xm(o) 8.1.3
Cy effect; Xm dependent cap.
i/Cy IAQy dt + Py(o) 8.1.4
of y side of actuator PY
Cz effect; Xm dependent cap. Pz
i/Cz IAQz dt + Pz(o) 8.1.5
of z side of actuator
Rsy effect; Xv dependent Qsy 0 if Xv is 0 I
Qsy = f (APsy) Kv " Xv " ~Psy if Xv is positive 8.1.6
No saturation of port opening 0 if Xv is negative)
Rsz effect Qsz 0 if Xv is 0 1
0 if Xv is positive 8.1.7
KV IXv I APsz if Xv is negative)
Rye e ffe ct Q~e 0 if Xv is 0 1
0Kv IXv I APye~ if Xv is positive 8.1.8
- if Xv is negative)
Rze effect Oze 0 if Xv is 0 }
Kv Xv dPze ~ if Xv is positive 8.1.9
0 if Xv is negative
Load friction, Rf effect Ff Sign ~ F~ + Kf 8.1 .i0
Power transformation, TF(Ay) Qya Xm . A y 8.1 .ii
Fy Py Ay 8 .i .12
Power trans formation, TF (Az) ~a ~ ' Az 8.1.13
Fz Pz Az 8.1.14
Supply 0-junction es Qsy + Qsz 8.1.15
Rsy 1-junction APsy SPs - Py 8.1.16
Qy O-junction QY Qsy- Qye 8.1.17
Rye 1-junction APye Py- Pe 8.1.18
Exhaust 0-junction Qe Oye + Qze 8.1.19
Rze l-junctlon AP ze Pz - Pe 8.1.20

Qz 0-junction Qz = - Osz + Qze 8.1.21
RSZ l-junctlon APsz = SPs- Pz 8.1.22
Cy 0-junction ~QY = Qy - Qya 8.1.23
Cz 0-junction AQz = Qza - Qz 8.1.24
Fa l-junction Fa = Fy - Fz 8.1.25
Fm 1-junction Fm = Fa - Ff 8.1.26
Auxiliary Equations:
Load displacement Xm(t) Xm = ]Xm dt + Xm(o) 8.1.27
Vy(o) is initial volume,
Cy = (vy(o) + ~y xm)/~ 8.1.28
Y side, B is bulk modulus
Vz(o) is initial volume,
cz = (vz(o) Az - ~I/B 8.1.29
Z side, B is bulk modulus
Xv = input to be specified 8.1.30
158
a. The System
Pressure
Com~n
st~pated I'l ~lI
] 0 ~Le ver System
Xi, Xv, Xm shown in their positivedirections
b. Simplified System
supply Oil Xi
ConstantPressure
! !
., I I
Ck
Xm
Rf
Co The P r o p o s e d Bond Graph
Supply Control Valve Actuator Load
/ ~ / \
~------Xi, ~ Py
~ o-.~ APye
~e/~'
I
Qs--~0sPs
$7~~-~ 7-,_t,......~n ~ -~
~I //}S Fz/F!e Fk~

Qsz QzeI Xi, Xm
\---Xi, >~ Qz Pz
FIG. 8.2 BOND GRAPH FOR LOADED HYDRAULIC SERVOSYSTEM
159
a pressure relief valve

a reservoir and its ancillary equipment
Many systems also include a high pressure filter near the pump discharge. Some
include an accumulator. It is reasonable to think of a power unit as being versatile
in that it could be used to provide power to any of a number of actuation systems.
Hence, for the system design situation it is attractive to have on hand dynamic models
for typical power supply sub-systems.
Fig. 8.3a shows one typical pumping system. The variable restriction is
included to represent the impedance load the sub-system will "see" when it is
connected to an actuator sub-system. The intention now is to prepare a fairly
detailed model which could be useful in a variety of applications of the pumping
system. Such a model could also be used to study the internal dynamics of the
pumping system itself.
The system can be conveniently divided into the three sections
the pump drive sub-system (electric motor, shafting, coupling)

the pump
the downstream components (relief valve, line, filter, valve).
Previous Figs. 5.4, 5.8, 7.9 developed a bond graph for an induction electric motor,
including its output shaft compliance. Fig. 7.9 is repeated as Fig. 8.3b. The
coupling and pump shaft unit can be included by lumping its inertia with rotor inertia
Ix , its friction with Rf and its compliance with Ck . From previous Fig. 7.11
the bond graph of Fig. 8.3c can be proposed for the pump. Fig. 8.3d shows a bond
graph encompassing the relief valve (Rr) , the resistance of the line plus filter
{Rb) , the capacitance of the line plus filter (Cb) , and the resistance of the load
valve (Rv) .
Fig. 8.3e shows the system bond graph formed simply by coupling the bond graphs
of Figs. 8.3b, c and d. Adjacent 1-junctions and adjacent 0-junctions have been
combined. The direction half-arrows and the causal bars were readily added using
physical sense and causal preferences. Combining the component sub-systems did not
upset their power flow directions and causalities.
The system equations written using the bond graph are given in Table 8.2. The
terms and symbols are as previously used.
Linear relationships were assumed for most of the dynamic effects in the model.
This assumption allows effects such as Ix , Ip , Ck , Cb , Rm , Rp , Rr ,
Rb , and Rv to be taken as linear coefficients. This is by no means always
160
necessary or desirable. It is obvious enough that Ix and Ip are coefficients--

the respective inertia values. Shaft compliance Ck is reasonably a coefficient.
The capacitance Cb is associated in the present example with fixed volumes, and
consequently the assumptions of linearity is reasonable. Linearity of Rm is
reasonable over a fairly wide range of electric motor performance. Leakage flew
through the clearances of the pump's working parts is viscous, and hence linear with
pressure over wide ranges of pressure. Assumption that the resistance Rb of the
filter and hydraulic line is linear is more suspect if we expect the model to give
large scale dynamic response. However, the pressure drop associated with Rb is
quite small relative to the pressure drop likely over the adjacent load valve Rv .
Consequently we probably need not describe Rb with much accuracy.
The overall load pressure drop, represented by APv across resistance Rv , is

undoubtedly large, and consequently we describe it by a normal square law orifice
flow relationship (equation 8.3.13).
The friction equations used for Tf and Tfp were discussed in Chapter 7.
Coulomb friction could be added.
The TF equations are the well known ideal relationships
pump torque = volumetric displacement per radian (Vp) - pump pressure (Pp)
pump flow = volumetric displacement per radian (Vp) pump speed (up)
For a variable displacement pump, volumetric displacement per radian of rotation pez
radian of swashplate angle ( V ~ ) multiplied by the swashplate angle (8p) in radi~s
gives the required pump volumetric displacement (equation 8.3.24).
The model, equations 8.3.1 to 25, was applied to a specific system, which was
also built and tested experimentally (Refs. 8.1, 8.2). A fixed restriction Rv was
used, and the system was activated for dynamic response by moving swashplate angle
8p from one position to another, increasing the flowrate demand of the system.
Some of the coefficients needed for the model were easily obtained, others
required experimental investigation or estimation. The values and how they were
obtained are indicated in Table 8.3.
Fig. 8.4 shows the results of a simulation of the model effected using CSMP on a
B6700 computer. Fig. 8.4a shows the input used, a change in swashplate angle.
Fig. 8.4b shows the consequent development of system pressure Pv(t) , together with
a corresponding experimental measurement. Fig. 8.4c shows simulation results using
a variety of values for the bulk modulus of the oil. Once a simulation is working
it is a simple matter to make this sort of excursion to investigate the sensitivity
161
a. A Pumping System
Variable Delivery Pump Restriction
Ele Ill I
~~ PressurFil~iief Val~e I .... I
b. Bond Graph for Induction Motor and Output Shafting

Re Rf Ix Ck
~-~-J" 0 - - - - ' ~ I ~' ": I I-----'- 0 =p--~--~

T~ ~x 0Jx TO
C, Bond Graph for Positive Displacement Pump

Rfp Rp
TfPI T Qp
o ='I: ~ " ~ ~ 0 : ~ .
Ty~ Vp pp
Ip
d. Bond Graph for Downstream Components

Rr Rb Cb Rv
~Qr ~ I T
~--~0:
Pp
"I,
Qpy
~0
Pv
'*:1: ~
Qv
'=
e. Bond Graph for Pumping System

R~ Rf Ck Rfp Rp Rb Rv
~-~0--~I~-L0--~I~ ~ ~K-L0~--~I~-~0----~I: ~
Ix Ip Rr Cb
FIG. 8.3 ~OND GRAPH FOR PUMPING SYSTEM (EXAMPLE 8.3)

162
TABLE 8.2 p o w e r State E q u a t i o n s for E x a m p l e 8.4
Component, Effect, Comment Equation

Sources :
E l e c t r i c m o t o r field s p e e d S~ = constant 8.3.1
Reservoir pressure Spe = 0 8.3.2
I effects:
Electric motor shaft s p e e d ex = i/Ix [TX d t + rex(o) 8.3.3
A
Pump input speed up = i/Ip ITy d t + ep(o) 8.3.4
C effects:
Electric motor output torque To = i/Ck Id~k d t + To(o) 8.3.5
System discharge pressure Pv = i/Cb IAQb d t + Rv(o) 8.3.6

R effects:
M o t o r torque, Re effect T~ = Rm A~ 8.3.7
M o t o r w i n d a g e p l u s m o t o r shaft
Tf = Kfl ~ x 3 + Kf2 - ex 8.3.8
friction, Rf effect
P u m p friction, Rfp effect Tfp = Kf3 - up 8.3.9
P u m p leakage, REp effect Qgp = 1/Rp Pp 8.3 .I0
R e l i e f valve, Rr effect Qr = 0 if Pp ~ P s e t
8.3.11
= i/Rr (Pp - Pset) if Pp > P s e t
Line p r e s s u r e drop, Rb effect dPb = Rb - Qpy 8.3.12
D i s c h a r g e flowrate, Rv effect = i/~v - ~ P v ~ 8.3.13
TF:
P u m p i n p u t torque Tp = Vp " Pp 8.3.14
Pump geometric discharge rate Q~ = Vp up 8.3.15
Junctions : d~ = S~ - ex (motor speed slip) 8.3.16
Tx = Te- Tf- To 8.3.17
Awk = ~ x - ~p (shaft torsional deflection) 8.3.18
Ty = To - Tfp - Tp 8.3.19
.Ol~y = Qp - Q p - ov 8.3.20
Pp = Pv + APb 8.3.21
dQb = Q p y - Qv ( c o m p r e s s i b i l i t y component) 8.3.22
APv = P v - Spe 8.3.23
Auxiliary Equations: Vp = V ~ - 8p 8.3.24
8p = controlling input, swashplate angle 8.3.25
TABLE 8.3 Coefficients for E x a m p l e 8.4
Coefficient Value Dimension Source

Sw = 157 rad/s 1500 r p m syn. field s p e e d (manufacturer)
Ix = 16.5 E-3 kg m 2 rotor i n e r t i a (manufacturer) p l u s s h a f t
a n d c o u p l i n g i n e r t i a (evaluated)
Ip = 1.15 E-3 kg - m 2 p u m p r o t o r i n e r t i a (manufacturer)
= I0 E-6 r a d / ( N - m) l e n g t h / ( M o f I - s h e a r mad.) o f s h a f t
ab = 3.3 E-12 m3/pa Vol. 4 . 5 2 E - 3 m 3 / b u l k m o d . 1380 M P a
R ~ = 53 N m/(rad/s) s l i p c o e f f i c i e n t (manufacturer)
Kfl = 0.316 E-3 N m/(rad/s) 3 e s t i m a t e d r o t o r s p e e d f r i c t i o n coefficient
Kf2 = 3.2 E-9 N m/Pa e s t i m a t e d s h a f t v i s c o u s f r i c t i o n coef.
Kf3 = 5 E-9 N m/Pa e s t i m a t e d p u m p rotor f r i c t i o n coefficient
~p = 0.5 El2 Pa/m3/s) p u m p leakage c o e f f i c i e n t (manufacturer)
Px = 4 .i E9 Pa/(m3/s) r e l i e f v a l v e c h a r a c t e r i s t i c (manufacturer)
Pset = 28 E6 Pa analyst's decision
= 0.91 E9 Pa/m3/s) manufacturer
Rv = 8.9 E9 Pa~/m3/s) from e x p e r i m e n t a l APv - Qv p l o t
V~ = 19.7 E-6 (m 3 / r a d ) / r a d pt~np m a n u f a c t u r e r ' s d a t a
ao Input Applied
ep, r adian
o.2!
/
Iml
0.i
time
I I I : -~
0.i 0.2 0.3 second
b. Response of Pv
Pv, MPa
2O
15
I0
Experimental
Initial Simulation
time
. i i I I
0.2 0.4 0.6 0.8 1 second
C. Effect of Bulk Modulus on Response

Pv, MPa
lO
J - - B = 1380 MPa
5 / / S/ .... B = 1035 MPa
0 ~ { / I I ?' BI= 518 Mia 2ecnd
0.2 0.4 0.6 0.8 1
FIG. 8.4 SIMULATION OF MODEL FOR EXAMPLE 8.3

164
of a system to changes in s p e c i f i c p a r a m e t e r s . The d y n a m i c d e v e l o p m e n t o f all system

v a r i a b l e s w a s available.
It is p r o b a b l e that if it w a s r e q u i r e d o n l y to study Pv(t) as 8p is varied,

the b o n d graph and its e q u a t i o n set could be s i m p l i f i e d greatly. Pt~np speed could
p r o b a b l y be taken as a source, allowing dismissal of the p u m p d r i v e ' s d y n a m i c s and of
p u m p friction and inertia. It is p r o b a b l e that line resistance Rb also could be
i g n o r e d as small c o m p a r e d w i t h Rv . H o w e v e r , i f d e t a i l e d study o f the p u m p u n i t
w a s required, the d e t a i l e d m o d e l should be used.
8.5 PUMP-CONTROLLED HYDROSTATIC DRIVE
Fig. 8.5a i l l u s t r a t e s the system to b e simulated. It is a s s u m e d that
the speed of the input drive is c o n s t a n t
load speed is c o n t r o l l e d b y e x t e r n a l a d j u s t m e n t xi of the p u m p stroke

control l e v e r
p u m p d i s p l a c e m e n t can b e zero or p o s i t i v e o n l y
the h y d r a u l i c c i r c u i t remains filled so that there is no cavitation (make-up

c o m p o n e n t s not shown)
the r e l i e f valve will n o t n o r m a l l y lift, b e i n g p u r e l y a safety device.
A v e r s i o n o f this system w a s u s e d as an e x ~ n p l e for c o n v e n t i o n a l m o d e l l i n g in

C h a p t e r 3.
Fig. 8.5b shows a b o n d graph for the system. It u t i l i z e s a structure for the
p u m p m o d i f i e d from Fig. 7.11g b y n e g l e c t i n g p u m p f r i c t i o n (Rfp) and i n e r t i a (Ip) .
It i n c l u d e s d e l i v e r y line r e s i s t a n c e Rh and h y d r a u l i c m o t o r leakage Ra . It
n e g l e c t s the e f f e c t s o n d y n a m i c response of the r e t u r n line flow. Pump, d e l i v e r y
line, and m o t o r c a p a c i t a n c e s are l u m p e d as Cpha . The h y d r a u l i c m o t o r is regarded
as a f i x e d - m o d u l u s p o w e r transformer, w h i l e the d r i v e n a s s e m b l y is r e p r e s e n t e d as a
r o t a r y i n e r t i a load w i t h friction. The set of p o w e r state e q u a t i o n s p r e p a r e d using
the b o n d g r a p h is shown on Table 8.4, the nomenclature b e i n g i n c l u d e d w i t h Fig. 8.4.
The initial v a l u e s mj(o) , Pp(o) and the coefficients Stop , Ij , Vpha ,

B , KZp , Kr , Kh , Ka , Kfj , Va , Kp m u s t b e s p e c i f i e d before s i m u l a t i o n
can proceed. The s i m u l a t i o n then p r o v i d e s ~j (t) , Pp(t) , etc. for s p e c i f i c
inputs Xi (t) . The m o d e l and simulation can be e x t e n d e d to allow for the e f f e c t s
of the return line, for c r o s s - p o r t leakage, and for o v e r - c e n t r e p u m p action (reversible
load drive).
165
a. S c h e m a t i c of S y s t e m
S~op
b. Proposed Bond Graph
Rp Ra Rfj
~ Q E P / Q r a
I Qa TfJ1
Pp
Sup
Pa Va mj
/
/ Q AP
/
Xi Cpha
c. D e f i n i t i o n of Symbols
S~p constant angular velocity of Qpa flowrate into motor

input drive Qa leakage in motor
Tp torque developed in input drive Pa pressure at motor
Vp swept volume of pump per radian Qa swept volume per second of motor
rotation Va swept volt.he per radian of motor
Xi angular displacement of swashplate Ta torque developed by motor
Q~ swept volume of ptEnp per second Tfj friction torque in motor per load
Pp pressure developed by pump ~j angular speed of motor per load
Qp leakage in pump Tj torque accelerating load
Qr flowrate through relief valve Vpha volume under compression
AQ loss of flowrate due to oil B bulk modulus of oil
compression K coefficient for an R effect
dPh pressure drop in line
FIG. 8.5 PUMP C O N T R O L L E D H Y D R O S T A T I C DRIVE

TABLE 8.4 M o d e l for E x a m p l e 8.5
Effect Equation
P~er S t a t e Equations:
S u p p l y source Sago = constant 8.4.1
Driven inertia Ij ~j = i/Ij JTj at + ~j(o) 8.4.2
J
Hydraulic capacitance Cph a Pp = I/Cpha IAQ dt + Pp(o) 8.4.3

P u m p leakage REp Q~p -- K~p Pp 8.4.4
R e l i e f valve Rr Qr = 0 if Pp ~ P s e t 8.4.5
= Kr (Pp - Pset) if Pp > P s e t
Line r e s i s t a n c e Rh dPh = Kh Qpa 8.4.6
M o t o r leakage Ra Qa = K % a Pa 8.4.7
Load friction Rfj Tfj = Kfj ~j 8.4.8
Pump transformer Vp Tp = Vp - Pp 8.4.9
Qp = vp. s~p 8.4.10
Motor transformer Va Ta = Va - Pa 8.4.11
Qa = Va - ~j 8.4.12
Pump 0-junction AQ = Qp- QEp - Qr 8.4.13

Line 1-junction Rh Pa = Pp- ~ph 8.4.14
Motor 0 - j u n c t i o n Ra Qpa = QZa + Q a 8.4.15
Load 1-junction Tfj Tj = Ta- Tfj 8.4.16
Auxiliary Equations :
C a p a c i t a n c e of pump, hose
Cpha = Vpha/B 8.4.17
and actuator
P u m p displ a c e m e n t / r a d Vp = Kp Xi 8.4.18
Control i n p u t Xi = i n p u t to b e s p e c i f i e d 8.4.19
8.6 VALVE-CONTROLI~D HYDROSTATIC DRIVE
The system, i l l u s t r a t e d on Fig. 8.6a, was u s e d to d e m o n s t r a t e p o w e r p o r t

m o d e l l i n g in C h a p t e r 4. It is d e v e l o p e d from Ref. 8.3. Fig. 8.6b repeats the
p o w e r p o r t s c h e m a t i c so that the r e a d e r can e a s i l y s t u d y the r e l a t i o n s h i p s between
the system, its p o w e r p o r t schematic, and its b o n d g r a p h w h i c h is s h o w n o n Fig. 8.6c.
The b o n d g r a p h i m p l i e s a q u i t e d e t a i l e d d y n a m i c m o d e l , as it includes
a p r e s s u r e c o m p e n s a t e d p u m p w i t h its i n p u t d r i v e
a r e l i e f valve
a filter
a 4 - w a y d i r e c t i o n a l control valve
a c r o s s - p o r t r e l i e f valve system, w i t h m a k e - u p supply from th~ charge pump
a hydraulic motor
the d r i v e n rotary i n e r t i a load.
167
i. C1=~it Diagram (Ii Fig. 4.61)
p~ll~ ~p.~l~'d P~p wotor
, - .............. ]I ~g~.~ i_.~ ~"

I W-__..~ 1 'x/H-4 r~:;~-F~ :~ I~'
l-J ~, ~---, 1 t~;9 ili "~t " ,~,I T 1 1
I YI~' I"\ li v ~ Vi I
b. P ~ PO~t D i e , r a m (all F I ~ . 4*&b)
-[ TI
c. non4 Grsph
-,,T/ ~
,,
Lj
i -~,i b0 "I' ~
~ . . . . 'P I
## ~3 II" ( w )
[- ~,o ...... ~T~-~'~o ~L,, o'2 ~,~ ~.,~. ,;1"
L; O~
I/
O"
ill
\ .............. 1%_ _ - J %. . . . . . . . . . . . . . . . . / % . . . . . . . . . . . . . . . . . . /
p~l~ ~ , ~ / control c~n Port ~ e t sylz~ S~li'luZ~= ~ = ~ ~,la
FIG. 8.6 VALVE-CONTROLLED HYDROSTATIC DRIVE (EXAMPLE 8.6, TABLE 8.5)

168
The additional information and model development on a bond graph relative to that
on a power port schematic is clearly illustrated. Each has a block representing the
system's c~mponents or sub-systems. Each shows the main power flow bonds. But that
is all that the power port schematic shows, while the bond graph goes on to expose
those effects in each block which are necessary, or that the analyst decides are
significant, to the system's dynamic performance. Further, the decisions on
causality provide both a technique for checking the appropriateness of a developing
model, and a framework for arrangement of the equations. The full arrows on the
power port diagram and the half arrows on the bond graph both indicate the presumed
directions of power flow in the system, and provide equation sign conventions.
The common effort at an 0-junction or flow at a 1-junction is shown adjacent to

the junction so that it need not be repeated on each bond of the particular junction.
Information on the present bond graph has been further advanced by inclusion of
graphic symbols indicating the nature of the non-linear relationships present. Thus,
the symbol at the pump MTF shows clearly that the pump is pressure compensated.
The graphic symbols are shown connected to the bond graph by active bonds (broken
lines, full arrowheads). Note the feedback from the pump discharge side 0-junction
to the pressure compensator, shown by an active bond. The 4-way control valve bond
graph allows only for positive load rotation. This decision reduces the valve's
bond graph from that of Figs. 8.1 and 8.2 to the two 1-junction resistances Rsy and
Rze , as Rsz and Rye are not allowed for. If required, full hi-directional
valve (and hence load) actuation can be reinstated.
Equations for most of the system's components have been presented in previous
examples and will not be repeated here. The cross-port relief block has not been
detailed previously. Using the bond graph its actions can be described by the
equations of Table 8.5.
The mathematical description, comprised of equations 8.5.1 to 19, is quite

extensive. But so too is the cross-port relief system itself, comprising of four
non-return (check) valves, two pressure relief valves and the various interconnections
between these and with the system itself. The bond graph and the equations given
allow that the cross-port'relief system might have to work in either direction of
motion of the hydraulic motor and load. The bond graph structure of the control
valve on Fig. 8.6c allows that the motor-load rotation will be in one direction only.
However, even with this restriction on control valve actuation, the cross-port relief
system may have to act in either of its directions.
16g
T A B L E 8.5 Equations for C r o s s - P o r t R e l i e f B l o c k , E x a m p l e 8.6
Effect Equation
F l o w t h r o u g h check v a l v e Qyl = 0 if &Pl p o s i t i v e ~ 8.5.1
R1 from Y line, and = K1 API if AP1 n e g a t i v e J
the a s s o c i a t e d 0- a n d
API = Py - Prl 8.5.2
1-junctions
Qy = Q y - Q y l + Qy3 8.5.3
F l o w t h r o u g h check v a l v e Qz = Q z 2 + Qze - Qz4 8.5.4

R2 fro~ Z line, and Qz2 = 0 if Pz < Prl
the a s s o c i a t e d 0- a n d 8.5.5
1-junctions = K2 AP2 if Pz > Prl J
~P2 = Pz - Prl 8.5.6
Pressure development and Prl = i/Crl [AQI d t + Prl(o) 8.5.7

flow downstream of J
check valves 1 and 2 AQI = Qyl + Qz2 - Qrl 8.5.8
into capacitance Crl
Flow through high Qrl = 0 if Prl < Pl set ~ 8.5.9

pressure relief valve = Krl (Prl - P1 set) if Prl > P 1 set J
Rrl
APrl = Prl - S P c 8.5.10
C h a r g i n g supply; assumed SPc = constant = P2 set 8.5. ll

to be at c o n s t a n t
p r e s s u r e and f l o w r a t e , SQc = constant 8.5.12
sPc and SQc ; n o Qr2 = S Q c - Q r l - Qy3 - Qz4 8.5.13
n e e d then to i n c l u d e
APt2 = SPc - SPe = constant 8.5.14
e q u a t i o n s for r e l i e f
valve Rr2 SPe = constant 8.5.15
Flow through check valve Qy3 = 0 if P y ~ pc ~ 8.5.16

R3 into Y line, a n d = K3 - &P3 if Py < Pc )
the a s s o c i a t e d I- a n d
AP3 = SPc - Py 8.5.17
0-junctions
F l o w t h r o u g h check v a l v e Qz4 = 0 if p z ~ pc
8.5.18
R4 into Z line, a n d = K4 . AP4 if Pz < Pc J
the a s s o c i a t e d i- a n d
AP4 = sPc - Pz 8.5.19
0- j u n c t i o n s
8.7 A LIFTING SYSTEM
8.7.1 Introduction
The r e q u i r e m e n t w a s to p o s i t i o n a load in a v e r t i c a l arc, using a hydraulic
c y l i n d e r to a c t u a t e a p i v o t e d beam, as i l l u s t r a t e d o n Fig. 8.7a. The s p e c i f i c a t i o n s
were
170
the load is 125 kg;
full load travel is 0.6 m;
control is to be manual;
the system is to be capable of causing the load to change position by 0.3 m

in 1 second;
load acceleration following the input command, and its retardation in

coming to rest, should each occupy about 15% of the stroking time of the
cylinder;
thus, the fastest required response of the system is as illustrated on

Fig. 8.7b;
for economy, the smallest pump and cylinder should be sought;
standard hydraulic components are to be used;
maximum system pressure allowed is 13.8 MPa.
Fig. 8.7c shows the main components of the hydraulic system proposed. It was
decided to use an open-centre control valve. The high gain between valve motion and
control flow in the valve's central position allows good control of load speed with-
out the need for constant pressure supply. The counterbalance valve in the lower
control line is needed to hold the load against gravity. The pressure relief valve
in the upper control line is used to limit cylinder pressure (and hence force and
acceleration) when the load is to be driven downwards, gravity providing much of the
force needed for the downward stroke.
Conventional static design criteria were used
to design the beam - - structural stiffness was the main consideration;
to select the pump and cylinder ....... this was associated with the selection
of beam lever ratio, with requirement to use system pressure maximum of
13.8 MPa, and with the desire to use the smallest pump and cylinder which
would provide the speed of load motion required (overall 0.6 m/s) ;
to select the valves.
The results were
beam length ~from pivot to load centre) 1.52 m
bean lever ratio (ratio load to actuator displacement) 5.1
effective total inertia of driven system (load 125 kg 141' kg

plus actuator moving parts and beam inertias referred
171
a. Task Requirement b. ResponseRequirement

j,Xm
Beam
,,
(
~ i Actuato~
y~aullc
0.15 O.B5 1 o
/ f
c. PrOl~Sed System MPa i "6m
..... ~_. ? .....

, __ Ill
ELI/
~13.8MPa~ Cont~oI
- Va/ve
ntxe
d. Bond Graph
C~
~'
~Y ......
I
h0 "~' "
LS ~
3 ~ ~ = ~-~0 ~-~-~0 ~ " 0 ~
~z sFg
T "
~Az TM
FIG. 8.7 LOAD L I F T I N G SYSTF~4 (EXAMPLE 8.7, TABLES 8.6 AND 8.7)
172
to the load, 16 kg)
main system pressure relief valve setting (as specified) 13.8 MPa
counterbalance valve pressure setting (required to 6.15 MPa

support load)
downstroke relief valve setting (required to limit speed 6.9 MPa

of downstroke)
hydraulic actuator bore 3.8 cm

s trok e 0.33 m
rod diameter 2.54 ~m
hydraulic pump, fixed displacement 9.8 x 10 -6 m3/rev
control valve 12 /min, 14 MPa
This brought the system design to the stage where a reasonably accurate assessment of
its dynamic response was needed to see if it satisfied the speed of response, and the
acceleration limiting specifications.
8.6.2 Develol~ment of Bond Graph
Fig. 8.7d shows a bond graph for the system developed on the bases that
pump drive speed is constant
positive valve action xv causes lowering of the load (pos. Xm )
the resistance of the short hydraulic lines have a negligible effect on load
position response
the capacitances of the lines can be lumped with Cp for the pump-valve
lines, and with Cy and Cz for the valve-actuator lines
the resistance of the filter is negligible, and its capacitance can be

lumped into Cp
the beam can be assumed to supply a constant lever ratio between actuator
displacement and vertical displacement of the load Xm .
Component bond graph structures utilized are similar to those developed in

Chapter 7 and used in previous examples. The driven load is modified to allow for
the lever ratio effect. Power flow directions are shown as positive to the Y side
and from the Z side of the actuator.
173
8.7.3 The Equations and Coefficients
The main purpose of the model is to allow prediction of the time response motion
of the driven load following cca~mand inputs applied to the control valve. The bond
graph of Fig. 8.7d allows development of equations for modelling the system for both
lifting and lowering of the load. AS seen in Section 7.9.5, many equations are
required for comprehensive modelling of the open-centre valve. In the present
system, lowering of the load is greatly assisted by gravity with little hydraulic
power requirement, the main need being to control the speed of descent. Thus
lifting of the load with hydraulic power is the main dynamic aspect. The equations
of the model are given in Table 8.6. They are written for lifting only.
The coefficients of the model are shown in Table 8.7.
8.7.4 Simulation
Fig. 8.8 shows the input and the responses of some of the system variables when
the model was simulated with MIMIC on a CDC Cyber 6400 computer. Load displacement
Xm(t) is directly related to piston displacement Xa(t) by the beam lever ratio.
Corresponding results obtained from the hardware system, which was assembled
8u]~sequent to the simulation, are included on Fig. 8.8. Considering the ambiguity
of some model parameter coefficients, the correspondence is good. It is a simple
matter to improve the correspondence between simulated and actual system performance
by adjustment of some of the model's parameters.
Comparison between Figs. 8.7b and 8.8b shows that the specified response is
approached but not achieved. The main discrepancy is the hump near t = 0.5 s .
Some parameter adjustment, possibly using optimization procedures of the type to be
discussed in Chapter 9, is called for. with the simulation working, it is a simple
matter to study the effect on response of adjusting a parameter of the model.
8.7.5 Conclusion
This example is developed more fully in Refs. 8.4 and 8.5.
8.8 A HILLY EfENAMIC ELECTROHYDRAULIC CONTROL SYSTEM
8.8.1 Introduction
Hydraulic control systems are commonly used where the precise positioning of
large resisting loads is required. High pressure hydraulics has little logic
capability, being a power transfer technology. Where logic or feedback is required
the hydraulic power system usually has to be interfaced with manual, mechanical,
pneumatic, electronic, or electrical control logic. Electrohydraulic servo-systems,
174
TABLE 8.6 Model for E x a m p l e 8.7

Hydraulic Supply
Speed source Swp = constant 8.7 .i
f
Pump pressure, Cp effect Pp = i/Cp - ]AQp d t + Pp(o) 8.7.2
Internal leakage, Rp
QZp = KZp - Pp 8.7.3
effect
Relief valve, Rr effect Qr = 0 if Pp ~ P s e t
8.7.4
= Kr - (Pp - Pset) if Pp > Pset
TF effect Qp = s~p vp 8.7.5
Tp = Pp vp 8.7.6
Pump 0-junction AQp = Qp - Qp - Qr - Qs 8.7.7
Control Valve
F o r all Xv , 0 to -+ x v m a x
Xv is spool d i s p l a c e m e n t
f r o m centre. ~ = Qsy + Qsz 8.7.8
Xv max is t h a t s p o o l Qe = Qye + Qze 8.7.9

d i s p l a c e m e n t w h i c h just
Qsy - Qye 8.7.10
fully opens valve. It QY =
is a s s u m e d t h a t Xv Qz = Qsz- Qze 8.7.11
cannot exceed Xv max .
Qsy = By (x(o) + xv) . ~psy % 8.7.12
X(o) is unde rlap.
Qye = ~ye (xCo~ - xv) apse 8.7.13
It is a l l o w e d t h a t K
Xsz - (x(o) -xv) - ~Psz ~ 8.7.14
values may be different. Qsz =
Kze ~X(o) + Xv~ - APze ~ 8.7.15
Note t h a t there are Qze =
Ps - P y 8.7.16
4 0-junction equations 6psy =
4 R equations APye = P y - Pe 8.7.17
4 1-junction equations 6Psz = Ps - Pz 8.7.18
i n the b a s i c m o d e l , one APze = 8.7.19
P z - Pe
for e a c h p h y s i c a l e f f e c t
on the b o n d graph.
A s s u m e s o p e n i n g s e q u a l for For X v = 0 only, i n a d d i t i o n to

all four o r i f i c e s w h e n e q u a t i o n s 8 . 7 . 8 to 19
spool centred, i.e. 8.7.20
APsy = Apye = APsz = APze = (Pp - Pe)/2
e q u a l underlaps.
When Xv ~ X(o) , p o r t s For positive X v ~ X(o) , in a d d i t i o n to
YE and SZ are closed. e q u a t i o n s 8.7.8 to 19
Qye = Qsz = 0 8.7.21
When Inegative X v I ~ X(o), F o r Inegative X v I ~ X(o) , in a d d i t i o n to

ports SY and ZE are e q u a t i o n s 8 . 7 . 8 to 1 9
closed.
Qsy = Qze = 0 8.7.22
Y Side, Line and Relief

Valve and Actuator
pressure development,
Py = I/Cy I~Qy d t + Py(o) 8.7.23
Cy effect
R e l i e f valve, Rry effect Qry = 0 if Py ~< P y s e t L
= Kry (Py - Py set) if Py > P y set_] 8.7.24
0-junction AQy = Qy - Q a y - Qry 8.7.25

175
TABLE 8.6 (contd.)

TF (Ay) effect Qay = Xa - Ay 8.7.26
Fy = Py Ay 8.7.27
Z Side. Line and Actuator

o~d Counterbalance Valve
If Xv is p o s i t i v e
r
Czl is c a p a c i t a n c e Pzl = I/czl IdQzl d t + Pzl(o) 8.7.28
b e t w e e n a c t u a t o r and J
c o u n t e r b a l a n c e valve.
Pzl(o) is the s e t v a l u e
o f the c o u n t e r b a l a n c e
valve.
TF (Az) effect Fz = PzI Az 8.7.29
Qzl = Xa Az 8.7.30
Czl 0-junction fiQzl = Qzl - Qz2 8.7.31
Counterbalance valve Qz2 = 0 if Pzl < PzI(o)

8.7.32
action, Rz effect = Qzl - AQzl if Pzl > PzI(o)
Rz 1-junction APz = Pzl - Pz 8.7.33

r
Cz2 is c a p a c i t a n c e Pz = I/Cz2 " IdQz2 d t + Pz(o) 8.7.34

between counterbalance J
valve and control valve.
Cz2 0-junction AQz2 = Q z 2 - Qz 8.7.35
Note i: W h e n the c o u n t e r -
b a l a n c e v a l v e opens, the
Rz effect becomes
negligible, Pzl and
Pz identical, and Czl
and Cz2 can be l u m p e d
as Cz .
Note 2: Equations 8.7.28
to 35 h o l d also for Xv
n e g a t i v e so l o n g as the
c o u n t e r b a l a n c e valve is
s e e n to b e e i t h e r full
open or full shut
d e p e n d i n g on the v a l u e
of Pzl and w h e n o p e n
h a s n e g l i g i b l e resist-
ance to flow.
Beo~ and Load System8

Net actuator force Fa = Fy - F z - F f 8.7.36
Friction, Rf effect Ff = s i g n X a F~ + K f Xa 8.7.37
Load motion, Im effect = i/Ira |Fro a t + 8.7.38
N e t force Fm = Fml + SFg 8.7.39
Gravitational force SFg = constant 8.7.40
Beam power transformation ~m/LR 8.7.41

TF (LR)
t~l = Fa/LR 8.7.42
176
TABLE 8.6 (contd.)
Component, Effect, Co~nent Equation
Auxiliary Equations
Xm d t + Xm(o) 8.7.43
Vy(o) is Y side volume Cy ~Vy(o) + Ay Xa)/By 8.7.44

under compression. By
is Y side bulk modulus.
vzl(o) is v o l u m e u n d e r Czl {Vzl(o) + Az xa]/Szl 8.7.45
compression between
actuator and counter-
balance valve. Bzl
is r e l a t e d b u l k m o d u l u s .
TABLE 8.7 Coefficients for Example 8.7
Power Supply Sup = 151 r / s

vp = 1 . 5 6 E-6 m 3 / r
Kp = 2 E-12 (mB/s)/Pa
Kr = 2.5 E - 9 ( m 3 / s ) / P a
Pset = 13.8 E6 Pa
Pp(o) = 0 . 5 E6 P a
cp = 2 . 9 1 E - 1 2 m3/Pa
Control Valve Ksy = Ksz = 1 0 . 5 E - 6 m2/(pa, ~ s)

Kye = Kze = 13.7 E-6 m2/(pa ~ s)
Pe = 0
x(o) = 0.019 m
Actuator, Y side Py(o) = 0.5 E6 P a

Kry = 2.5 E - 9 ( m 3 / s ) / P a
Py set = 1 3 . 8 E6 P a
Ay = 0.63 E-3 m 3
Actuator, Z side PzI(o) = 0.5 E6 P a

Az = 1.14 E-3 m 3
Cz2 = 0.2 E-12 m3/pa
PZ(O) = 0 . 5 E6 P a
Bee~ and Load F~ = 120 N

Kf = 1 . 5 7 E3 N s / m
Im = 141 kg
~m(o) = 0
SFg = 1 . 3 8 2 E3 N
LR = 5.1
Auxiliary Equations Xm(o) = 0
Vy(o) = 0.25 E-3 m 3

By = 380 E6 P a
V z l (o) = 0 . 2 2 E-3 m 3
Bzl = 5 3 5 E6 P a
177
a. Input xv(t)
o V
I ! I I ~ I t I I ~ Q a
0 i
b. Actuator Piston D i s p l a c e m e n t Xa(t)
0.15
0.1
Xa, m
measured
0.05
simulated ..+.........
f
- tim, s
0 1
C. Actuator Piston V e l o c i t y Xa(t)
Xa, m/s II " ~$ ~ I "a% I

0.25 I ~
time, s
0 1
d. Actuator Pressure Development Pzl(t)
14
Pzl, MPa
;:.IJ -- time, S
0 0.2 0.4 0.6 0.8 I
FIG. 8.8 RESPONSE OF LOAD LIFTING SYSTEM

178
long used in military and space technology, are now widely utilized in industrial
applications such as machine tool control, flight simulators, structure fatigue
testing rigs, computer-aided manufacture, and industrial robots. With increasing
use of microprocessors and computer control it can be expected that electrohydraulic
servovalves will be used increasingly in industrial and manufacturing automation.
Electrohydraulic drives are often used where the dynamic response requirements
of the driven load(s) in terms of speed, cycling, accuracy, and stability, are
particularly severe. One such application, the subjec~ of the present Example, is
the control and drive system associated with a 28 tonne seismic block facility.
The intention is that the electrohydraulic system shakes, at controlled force and
displacement amplitudes and frequencies, a structure or machine undergoing
vibrations analysis. The present system is intended to be useful to 300 Hz.
Fig. 8.9a illustrates the shaker system layout and Fig. 8.9b the essentials of the
hydraulic system. The hydraulic power supply centres on a 30 kW Denison piston
pump operating at up to 21 MPa set pressure and up to 60 /min flowrate.
A 30 kW induction motor drives the axial piston pump at i000 rpm. The main
pressure control valve is set at between 2.76 and 20.7 MPa, depending on require-
ments, to provide constant pressure at that value to the servovalve. The return
line pressure control valve is set at between 2.1 and 3.45 MPa, depending on the
system performance requirements, to prevent the possibility of cavitation with
consequent dynamic performance uncertainty. Both pressure control valves are
pilot operated by remote control from the servo-controller. The electrohydraulic
servovalve receives its position feedback modulated input signal from the servo-
controller. The water-cooled oil cooler enables oil temperature to be controlled,
thus controlling oil viscosity, an important factor in high performance hydraulics.
The long hydraulic hoses provide flexibility so that the actuator can be taken to
any desired position on or around the seismic block (approximate dimensions 3 m x
4.5 m area x 0.65 m depth 28 tonnes mass) . The accumulator package provides the
smoothing of pressure transients and the capacitive effects needed for smooth load
position control. The electrohydraulic system has various override, safety, and
functional features not included on Fig. 8.9 as they have little effect on dynamic
performance under normal system operation.
8.8.2 The Bond Graph
In order to better understand the dynamic nature and performance of the system
beyond the rather meagre information supplied by the manufacturer, a simulation of
the shaker system was required. Because the main interest lay in employing the
upper limits of dynamlc response, the model for simulation needed to be detailed,
accurate, and versatile.
179
a. ~hmlat~c ~a~t
+" +" I l
I,+'.I)Ut SI~nal'*---ll,,1- ~ } ~$I~II I . l
I - I
1 ] 'l
+.- . . . . . . . . . . I /
_/
b. Ellentlal# o f H y d ~ a u l i c Sylte~
CICU~K3~tI~I~
S*+~I++
. ]~:.z~. C,aph
/
mp
"+-/
F/ire:
\ 2-n . . . . . ,,
f
.1.. F~ 1' n....++./ Roe 1
\I
lllltxoh~*mL~+c T m l m l l l . In~ toId
Py
"\2 I +\,/
+T ? ? - i
r ' ~ l ~+'+'~ 0 - " * + I + - - ' ,
r~
1!_! + i~ / + , + +\
*/
]
0 +~+-i 1 t.,..-- 0 J.,,L~ I p.+.-- 0 }+..--+ O .,+--.i 1
p.iJr.) Itrm(R)
, ~ J+r,
X#
; j,J,~t +,+F+.'~+++++ j,
++
cooler
control
FIG. 8.9 AN E L E C T R O H Y D R A U L I C S Y S T E M (EXAMPLE 8.7)

180
Fig. 8.9c shows the power bond graph structure developed for the system, the
various hardware components being indicated adjacent to the appropriate sections of
the bond graph. Some of the R and C effects included will not be simple
coefficients. Discontinuous and other non-linearities are present.
The pump is represented as a constant flow source SQ (constant angular speed

and constant swept volume rate). Internal pump leakage is accounted for as R~p ,
and the resistive and capacitive effect of the pump's discharge side is recognized
as the combination Rb and Cb . Next is the main system pressure regulator,
included as effect Rrl. This regulator sets the system pressure. Proceeding
right along the main power flow line, the hydraulic accumulators and the hydraulic
hoses are allowed for as appropriate pairs of R and C effects.
The closed-centre electrohydraulic servovalve is of single-stage design. Its

internal leakage, flow forces, friction, spool mass, and torque motor characteristics
have all been neglected in the proposed bond graph. This is justified by the
present interest being centred on the performance of the actuator and driven load
and not on the servovalve itself. The bond graph for the servovalve is identical
to those used previously for closed-centre valves (Figs. 7.15, 8.1, 8.2). It is
necessary only to realize that Rsy , Rsz , Rye , and Rze are automatically and
continuously adjusted by the electrical signal from the servo-controller. This is
indicated on the bond graph by Rsy(H) etc. This input plus feedback situation
could have been indicated via signal (non-power) bonds on the bond graph leading to
each of Rsy , Rsz , Rye , and Rze , as illustrated on Fig. 8.10. Servo-
controller control of pressures Prl and Pr2 could similarly be indicated via
signal bonds from the servo-controller to Rrl and Rr2 respectively. However
these signal bonds clutter the bond graph. To a person familiar with the
technology, as any designer or analyst should be, and appreciative of the relation-
ships between Figs. 8.9a, b, and c, the modulation of the R effects noted should
be obvious without the inclusion of the signal bonds, and even without use of the
R(H) or R (input and feedback effects) notation.
The actuator and driven load sections of the bond graph are similar to those
used in a number of earlier examples. The return line bond graph includes
recognition of its hoses, its accumulator, and its pressure control valve (Rr2) .
The return line from the EHSV joins with the discharge from the supply line pressure
control (Rrl) , and the total discharge passes through the oil cooler to return to
tank. The capacitance and resistance effects of the oil cooler and its associated
lines are included on the bond graph as Cx and Rx .
181
To Rrl and Rr2 J ] Desired State

] Servo-controller and ~6
| signal modulator |
system ~_
AYE
I I !
I
Signals ---P"
I "0~
x'/
",0 01 Xa I ~ Im
~I:
I T I
"0 ~I' I
/
FIG. 8.10 DIRECT INCLUSION ON BOND GRAPH OF MODULATING SIGNALS
8.8.3 Conclusion
Further explanation of bond graph development, the equations derived using the
bond graph, and results of a simulation of the system are given in Refs. 8.6 and 8.7.
There are 77 equations in the model, most of forms already used in this Text. It
should not be thought that the system bond graph of Fig. 8.9c came easily, without
effort. To form and link the component bond graph structures took considerable
effort in satisfying physical reality, adequate completeness, and causality.
However, development of the bond graph provided the key to orderly and progressive
model development in a situation which would have been daunting if approached by
conventional modelling procedures.
8.9 CONCLUSION
A number of examples has been given of the use of bond graphs to provide a
link between a hydraulic control system, real or proposed, and its simulation for
analysis and synthesis purposes.
182
8.9 REFERENCES
8.1 Winton, R.D., An Investigation into the Mathematical Modelling of

Hydraulic Control Systems, Using Power Bond Graph, Project Report,
Dept. Mech. Eng., Monash University, Australia, 1973.
8.2 Dransfield, P. and Winton, R.D., Applying Power Bond Graphs to Hydraulic
Control Systems, Proc. III Congresso Brasiliero de Engenharia Mecanica,
Rio de Janeiro, 1975.
8.3 Stecki, J.S. and Cuthbertson, S.C., already cited as Ref. 4.3.
8.4 Barnard, B.W., The Dynamic Response of a Hydraulic System, M.Eng.Sco

8.5 Barnard, B.W. and Dransfield, P., already cited as Ref. 7.4.
8.6 Teo, M.K., Development of a Power Bond Graph Model of an Electrohydraulic

vibrator System, Project Report, Dept. Mech. Eng., Monash University,
Australia, 1974.
8.7 Dransfield, P. and Teo, M.K., Using Bond Graphs in Simulating an

Electrohydraulic System, J. Franklin Inst., V.303, N.3, Pergamon, 1979.
CHAPTER 9
OPTtMISING SYSTEM RESPONSE
9.1 INTRODUCTION
The ethos of this Text is that ensuring good dynamic response should be an
integral part of the design of a proposed hydraulic control system. Further, that
the securing of this aim is b e s t effected via mathematical modelling of the proposed
system and digital simulation of the model. Further still that bond graph techniques
provide a dynamic modelling procedure particularly suited to the modelling of hydraulic
control systems.
To here the Text has concentrated on developing the bond graph procedures needed
to model hydraulic control systems. It has been implicit that such models are best
simulated by using digital simulation.
Having obtained a predicted response questions arise as to its quality. As

illustrated in Chapter i, Fig. 1.3, a predicted response might be satisfactory,
promising, or hopeless. If promising, the designer-analyst should be able to develop
his proposal towards the satisfactory state. If hopeless, he may have to consider a
more drastic re-design to at least get model performance to a promising state.
In the present Text, optimization of response is regarded as the process of

causing a promising response to converge on a desired response by adjustment of some
model p a r ~ e t e r s . The model is optimized if its response shows the minimum deviation
from the desired response possible under the variations of parameters allowed or
practicable. Thus if a model is optimized it is not implied that its response has
converged fully on the desired response. Rather, that it has converged on it as
closely as is practicable. A system model and the allowed range of its parameters
184
m a y be incapable of producing the precise required response. An optimum response is
not necessarily acceptable.
As seen in the examples of Chapter 8, a system's model will have many coefficients.
To retain satisfactory power and motion characteristics m o s t of the coefficients will
not be able to b e varied much w i t h o u t basic re-design o f the system. However, there
are usually a few coefficients for which the designer-analyst has a reasonable choice
of values. The idea becomes to seek that combination of values for these parameters
which yields optimum response.
In the simplest case only one parameter is varied. A set of discrete solutions
of the system model, with the chosen parameter varied over its allowed range, quickly
leads to the optimum value of the parameter; that is, that value w h i c h produces the
desired system response, or the possible response closest to the desired response.
In the more likely case where two or more parameters can be varied simultaneously,
the number of possible combinations of values of the chosen parameters is so formidable
that computer techniques are required for the search for the optimal combination of
parameter values .....again, that cc~bination w h i c h provides response closest to that
desired.
This Chapter describes a response optimization scheme that has been developed
and used with hydraulic control system response (Refs. 9.1, 9.2). There is no
attempt to generally survey control system response optimization.
9.2 THE REQUIREMENTS
Requirements for the performance of analytical response optimization of the kind

envisaged axe :
i. Specification of a realistic response-inducing input.
2. Specification of the response desired.
3. Development of a mathematical model w h i c h describes the system's dynamic

performance and relates response to input.
4. Means to readily effect solutions of the model to provide predictions of

system response.
5. Means to compare desired {Xl(t)) and predicted (X2(t)) responses leading

to description of error as
E(t) = Xl(t) - X2(t)

185
6. A criterion for judging the o v e r a l l value or s i g n i f i c a n c e of the e r r o r function

E(t) such that m i n i m i z a t i o n o f the c r i t e r i o n m i n i m i z e s the e f f e c t o f the e r r o r
and m a x i m i z e s the q u a l i t y of the response relative to the d e s i r e d response.
7. A m u l t i p l e - p a r a m e t e r search p r o c e d u r e capable of e x p o s i n g that c o m b i n a t i o n of

values for the p a r a m e t e r s chosen for o p t i m i z a t i o n p u r p o s e s w h i c h p r o v i d e s o p t i m u m
response.
8. M e a n s w h e r e b y the d e s i g n e r - a n a l y s t can q u i c k l y and r e l i a b l y effect Step 3. It

h a s b e e n p r o p o s e d that p o w e r b o n d graph techniques provide this.
9. M e a n s w h e r e b y Steps 4 to 7 can b e m a n a g e d b y the d e s i g n e r - a n a l y s t w i t h a h i g h

degree o f automation. The digital c o m p u t e r p r o v i d e s this facility.
9.3 ERROR CRITERIA
Fig. 9.1 i l l u s t r a t e s a d e s i r e d response xl(t) and a p r e d i c t e d r e s p o n s e X2(t) .

Error E(t) is r e p r e s e n t e d b y the shaded area b e t w e e n the two curves. If the shaded
area can b e eliminated, we h a v e p e r f e c t optimization. If it can b e minimized, b y
u s i n g a l l o w e d v a r i a t i o n of s e l e c t e d p a r a m e t e r s o f the m o d e l to adjust X2(t) , the
response h a s b e e n optimized. Thus, the m o s t o b v i o u s e r r o r c r i t e r i o n is the m i n i m i z -
a t i o n of the total s u m m a t i o n o f error, i.e. m i n i m i z a t i o n of the shaded area b e t w e e n
Xl(t) and X2(t) . The e r r o r criterion is called mini2nization of absol~t8 va~u~ of
error. The shaded area c a n b e described
t
~E =
i o
li~ dt
where IAE denotes Integral of A b s o l u t e v a l u e of E r r o r

IEI denotes a b s o l u t e v a l u e o f error
t is the time over w h i c h the c o m p a r i s o n of xl(t) and X2(t) is made,
m e a s u r e d from the time (t = 0) w h e n the i n p u t is applied.
M i n i m i z a t i o n o f the error function IAE b e c o m e s the absolute error criterion.
M a n y v a r i a t i o n s in the b a s i c e r r o r function have b e e n d e v e l o p e d to b i a s or w e i g h t

the e r r o r m i n i m i z a t i o n p r o c e d u r e to b e sensitive to p a r t i c u l a r c h a r a c t e r i s t i c s of a
s y s t e m ' s r e s p o n s e potential. Table 9.1 shows a n u m b e r of error functions.
L a B r o o y [9.1] c o n s i d e r e d m o s t of these in r e l a t i o n to r e s p o n s e s t y p i c a l l y r e q u i r e d
of h y d r a u l i c control systems. It was confirmed o r c o n c l u d e d that
IAE is less sensitive to large e r r o r and m o r e sensitive to small e r r o r

relative to ISE and is therefore likely to p r o d u c e a n o p t i m u m w i t h larger
186
Response / ~ ~ Predicted response X2(t)
/ x
f l /
~11 \Desired
- ..~
r o r E{t)
response xl(t)
time
FIG. 9.1 ILLUSTRATING DYNAMIC ERROR
TABLE 9.1 Error F u n c t i o n s
Error Function Designation Definition
ISE
Its E " E dt
o
ItSE
I ts t E E dt
o
IStSE
I ts t t E E dt
o
z~
I ts IEI dt
o
ItAE
I ts t " IIE d t
o
ts 2Kt
IXSE
i o
e E E dt
IMXSE C[1 o
- e -2K E E dt
I d e n o t e s 'integral of'
S d e n o t e s ' square of'
E denotes error
t d e n o t e s time
A d e n o t e s 'absolute v a l u e of'
X d e n o t e s 'exponential w e i g h t i n g of'
M d e n o t e s 'modified v a l u e of'
ts is s i m u l a t i o n time (> s e t t l i n g time of m o d e l response)
e is the e x p o n e n t i a l
K is the a r b i t r a r y c o n s t a n t u s e d w i t h IXSE and IMXSE
187
rise time but less overshoot and oscillation,
ISE is heavily influenced by the initial stages of response, when error is

largest, but is insensitive to lingering oscillations of small amplitude, and
is likely to produce an optimum response with light damping, fast rise time,
considerable overshoot, and considerable oscillation,
time weighting (ItSE , ItAE , IStSE) reduces the contribution of initial

error and increases the contribution of prolonged error. This is likely to
encourage shorter settling time. However, it may inhibit fast initial
response,
the exponentially time-weighted functions (IXSE , IMXSE) , with appropriate

choice of K , allow a wide range of error function characteristics to be
obtained. These functions weight prolonged error exponentially and are
likely to yield optimum responses more heavily damped than do the other
functions. Judicious selection of K can begin the exponential weighting
at any point in the time frame of the response.
a Unimodal
Value of integral
b. Mul timodal error function
c. Non-modal
Value o~ parameter
FIG. 9.2 SOME ERROR FUNCTION SHAPES

188
Hydraulic control systems are stiff and are u s u a l l y h e a v i l y loaded. They are
expected to i n h i b i t f a s t r e s p o n s e to step i n p u t s for example, possibly with some
overshoot and o s c i l l a t i o n . It a p p e a r s that ItAE is suitable as an e r r o r function

in these circumstances, and i t is a d o p t e d for the example of this Chapter. However,
this s h o u l d n o t be i n t e r p r e t e d as e n d o r s e m e n t of ItAE to the g e n e r a l e x c l u s i o n of
o t h e r error functions.
9.4 SEARCH PROCEDURES
9.4 .i Introduction
The form of the value o f an error f u n c t i o n as o p t i m i z a t i o n parameters are changed

over their a l l o w e d r a n g e s can be
unimodal, possessing a minimum as i l l u s t r a t e d for single p a r a m e t e r
optimization on Fig. 9.2a,
multimodal, possessing several m i n i m a , Fig. 9.2b,
non-modal, possessing a minimum at one of the p a r a m e t e r boundaries,

Fig. 9.2c.
The u n i m o d a l case is the m o s t s t r a i g h t f o r w a r d . The m u l t i m o d a l situation can

cause p r o b l e m s , as a search p r o c e d u r e might direct itself towards a local rather than
the global minima. The n o n - m o d a l case a p p e a r s to b e unambiguous; however it can
cause computational p r o b l e m s as the search heads to the h a r d b o u n d a r y at w h i c h the

minima is located. Prior to p e r f o r m i n g a detailed error f u n c t i o n analysis, a
designer-analyst would rarely know which of the three s i t u a t i o n s o f Fig. 9.2 he has.
A good search p r o c e d u r e h a s to b e a b l e to seek a g l o b a l rather than a local minimum,

and has to be c a p a b l e o f n e g o t i a t i n g hard boundaries. Constrained limited multi-
parameter optimization is required.
9.4.2 Single P a r a m e t e r Optimization
Young [9.3] c o n s i d e r e d that the G o l d e n Section Search Procedure described by

Meichke [9.4] w a s e f f i c i e n t and e f f e c t i v e for i n c o r p o r a t i o n into the digital simulation
procedure MIMIC to p r o v i d e an a u t o m a t e d single-parameter optimization procedure
suitable for the class of systems r e l e v a n t to the p r e s e n t Text. Initial error
function evaluations at fixed interval over the range o f the c h o s e n o p t i m i z a t i o n
parameter can reveal the r e g i o n in w h i c h the true m i n i m u m exists. Multimodal minima

is u n l i k e l y to be a problem with hydraulic control systems as the r a n g e o v e r which a
parameter can b e v a r i e d realistically is l i k e l y to be quite narrow. It is m o r e

likely that the error function will be unimodal or non-modal.
189
9.4.3 Multi-parameter Optimization
The error function shapes illustrated on Fig. 9.2 apply to single-parameter

optimization. It can be seen that 1-parameter optimization required 2-dimensional
error function representation. For 2-parameter optimization, the lines of Fig. 9.2
become surfaces; i.e. the error function for 2-parameter optimization requires
3-dimensional representation. In general, an error function is related to the n
parameters of an n-parameter optimization scheme by a shape in (n + i) space.
When n is larger than 1 or 2, the shape becomes complex and abstract. Scanning
such a space for an optimum combination of the n parameters is a substantial
computational task. A manageable, efficient, and reliable search proceaure for
handling constrained and at~least 3-parameter optimization is required for hydraulic
control system analysis.
The Golden Section Search Procedure for 1-dimensional optimization does not
effectively extend to multi-parameter use [9.5]. More practical search techniques
are based on the evolutionary operation method of Box [9.6].
The Simplex method [9.7] is one EVOP approach. A Simplex is defined generally
as a shape formed by (n + 2) points distributed in (n + i) dimensional space, n
being the number of optimization parameters. Error function evaluations are made
for each of the (n + 2) points of the Simplex. The point producing the worst
evaluation, i.e. that point furthest from the point producing the smallest evaluation,
is discaxded and replaced by a point located in a more favourable position. This
point is located using a set of mechanical rules intended to help the Simplex converge
towards the point of minimum possible error evaluation. It has been reported [9.8]
that Simplex is most satisfactory when the error function is linear and when no
constraints are placed on the range of each optimization parameter. Also that
Simplex is susceptible to collapse into a sub-space to yield a local rather than the
global optimtun. For these reasons, Simplex seems to be inadequate for optimization
of the response of hydraulic control systems. It is recognized that Simplex
techniques might have been improved to accommodate its earlier limitations.
The Complex search method proposed by Box [9.9] and improved by Guin [9.10]
appears to be well suited to the present requirements. Complex is heuristic (self-
learning), is capable of handling non-linear error functions with unequally constrained
optimization parameters, can negotiate hard constraint boundaries, and can turn
corners in the space being scanned. It is unlikely to collapse into a sub-space to
yield a local minimum, as a single point existing in a "better" region (i.e. closer
to the global minimum) can cause the complex to be re-directed towards this point.
190
9.4.4 Complex
A general complex of dimension sigma is defined as a figure formed b y sigma

points distributed in the (n + I) space of the n-parameter optimization scheme.
Sigma must be ) (n + i) , and the optimization parameters explicit and independent
of each other. The initial distribution of the sigma points of the complex is
random except that at least one point must be within the optimization space specified,
i.e. it must satisfy all constraints on the chosen optimization parameters. If
other points lie outside the constraint boundaries, they are moved back to just
inside by formal Complex procedures.
The random nature of the casting of the initial complex of points allows a
probability that a point may be in the region of the global error function minimum
which locates the optimal combination of optimization parameters. In this case, the
Complex procedure will cause the complex to collapse towards the optimal point. If
no point of the initial complex lies in the region of optimality, the heuristic rules
governing the motion of the complex cause eventual scanning of the optimal region.
Like simplex, the rules of Complex generate new points to replace "worst" points with
the objective of collapsing the complex towards the global optimum point. Like most
search procedures, a wide choice of error function is allowed.
Fig. 9.3 illustrates, with a simple 2-parameter example, the nature of Complex.
The ordinates show the boundaries of the two optimization parameters N1 and N2 .
The small-chain broken lines illustrate contours of constant values of the chosen
error function as N1 and N2 are varied. Adopting sigma = 3 (i.e. at least
n + 1 ), the three points of the complex are shown as N1 , N2 , N3 , and their
centroid is at N . To minimize the value of error function, the complex must be
directed from the i0 contour towards the 4 contour. This is effected by reflecting
the worst point, N1 , through and beyond the centroid to NI' The next-stage
complex becomes N2 , N3 , NI' It can be seen that N3 would suffer the next
reflection through the new centroid.
Fig. 9.4 shows the flow chart developed b y La Brooy [9.1] for compiling the
optimization algorithm COMPLEX which he used with MIMIC to provide an automated
optimization procedure. The flow chart was modified from that of Box [9.9] for
increased efficiency. A system model simulated in MIMIC calls on the sub-program
COMPLEX which returns a parameter combination via multiple RETURN statements for
MIMIC to effect a response solution. MIMIC then evaluates the error function for
the particular parametric combination and recalls COMPLEX, passing the error function
value and receiving another parameter combination in return. The sole purpose of
the digital simulation procedure is to provide objective error function evaluations.
COMPLEX implements the logic of Fig. 9.4 to enable the Complex procedure to function.
It is a discrete Fortran optimization sub-program containing several parameters such
191
N2
N2max
%.,. __ .~, ~ ' ~ " NI' ,~/7
N2 / ~'
N2min I l- N1
Nlmin Nlmax
FIG. 9.3 REFLECTING A COMPLEX POINT
I Pick a feasible start point I

L Generate a complex point I
Relocate point
Check explicit Violated inside violated 1
constraints > boundary by a
Move point half- I distance delta
way towards I Not
centroid of Violated
remaining points ~p Has initial \
Violated / Check imp] Not vio]
plated [ complex been ~ No
\ constrair ~'~ generated?
~ Yes
I Ensure existence of I
objective function at
each complex point
#
Yes
< Check convergence >
Reflect the
lowest point No
through the No Is the low reflected > Program
centroid of point a repeater? failed
the remaining
points ~
Yes
I Halve alpha ]
Reflect lowest }
point using No Is alpha less than >
the new value theta?
of alpha
~ Yes
IS the second lowest Yes
point being reflected? /
Replace reflected point

with the point originally
unsuccessfully reflected
#
I Reset alpha to its
original value
Reflect this L Set 2rid lowest point I
'lowest' point ]~ to be the 'lowest point' f
FIG. 9.4 FLOW CHART FOR COMPLEX
192
as A L P H A and B E T A in Fig. 9.4 to w h i c h v a l u e s a p p r o p r i a t e to the p a r t i c u l a r problem
can b e ascribed. U s u a l l y one or two a t t e m p t s at o p t i m i z a t i o n are r e q u i r e d b e f o r e
knowledge adequate for a successful optimization is gained. Used initially with
MIMIC, COMPLEX was later c o m b i n e d w i t h CSMP.
9.5 EXAMPLE
Fig. 9 . 5 a and b i l l u s t r a t e a hydraulic servo-positioning s y s t e m a n d its b l o c k

diagram model [9.11]. Fig. 9 . 5 c shows the l o a d p o s i t i o n response required for u n i t
s t e p input. The two s e r v o s y s t e m coefficients K1 and K2 are a v a i l a b l e for system
response optimization purposes in the ranges
0.01 < K1 < 0.05

0.3 E-3 < K2 < 0.4 E-3
The e r r o r function ItAE and a six p o i n t complex w e r e c h o s e n and incorporated in the

COMPLEX-MIMIC package. One o f the initial complex points was chosen to be
K1 = 0.03 and K2 = 0.35 E-3 , mid-point in the a v a i l a b l e ranges. The r e m a i n i n g
five p o i n t s of the initial complex were randomly generated.
A f t e r 18 r e j e c t i o n s of "worst" points of the complex - - i.e. 18 r e f l e c t i o n s and

iterations -- error function values were within 30 E-6 of e a c h other. The
centroid of the final complex
K1 = 0.024
K2 = 0.3 E-3
was accepted as o p t i m i z a t i o n .
Fig. 9 . 5 d shows the r e s p o n s e of the o p t i m i z e d system. F o r comparison,

Fig. 9 . 5 e shows the r e s p o n s e s for K1 and K2 at their b o t t o m and top a l l o w e d
values respectively.
Note that the o p t i m u m v a l u e for K2 is at one of the K2 boundaries. It is

probable that an i m p r o v e d response would be achieved if the K2 boundary c o u l d be
lowered. The same e x a m p l e w i t h r e s p o n s e c r i t e r i a of s h o r t e s t rise time, m i n i m u m

overshoot, and m i n i m u m settling time following step input y i e l d e d the o p t i m u m
condition K1 = 0.0182 , K2 = 0.323 E-3 , a n d the r e s p o n s e s h o w n o f Fig. 9.5f. The
responses for K1 and K2 at their b o t t o m and top a l l o w e d v a l u e s are included.
9.6 C O N C L U S ION
A certain combination of e s t a b l i s h e d steps and p r o c e d u r e s which together form a

193
a. S e r v o sy s t e m b. Block Diagram of System

Xe xo
r r
,,, ~ ,
Co Desired Response of Xo to d. Optimum Response

Step Input of Xi
xo Xo
81 ptlmum
/W##~ desired 1
; ~ t SeCS , : ~ t secs
0 0.2 0.3 0 0.I 0.2 0.3
0.i
e R e s p o n s e s for K1 a n d K2 a t T o p
and Bottom of Their Ranges
xo
8!
4-
//
/
ldesired
/
l
o.1
~<i -- o . o i ,
l
0.2
K2 = o.oooa
~ t secs
o:a'-
f.
Xo
Optimum response
Response Sought
F
if F a s t e s t
bottom range response
4 //~///~/
I/// top range response
Xo
f" , Lt secs
s~ 0 011 0:2 02a-
K1 = 0.05, K2 = 0.0004
| .... | ; ~- t secs
0.i 0.2 O.3
FIG. 9.5 EXAMPLE

194
scheme for optimizing the response of hydraulic control systems has been described
and demonstrated. The component steps and procedures are:
specification of a desired response to a specified input, in form, time,

and magnitude
mathematical representation of the system, and selection frcm within that

model of several parameters for use in optimization
selection of an appropriate integral error function whose minimization

becomes the error criterion. ItAE is suggested for typical industrial
hydraulic control system responses
use of COMPLEX to search for the optimal combination of optimization

parameters
use of digital simulation to effect the repetitive model solutions required

for estimations of error.
The example presented was solved on a COMPLEX-MIMIC or COMPLEX-CSMP package

which integrated the steps and procedures required for optimization.
It is emphasized that optimization of the model's response is sought. The

system will be optimized only if the model truly represents the system.
There are many variations of optimization procedures available in the literature.
9.7 REFERENCES
9.1 LaBrooy, R., On the Optimisation of Hydraulic Control System Response,

M.Eng.Sc. Thesis, Monash University, Australia, 1975.
9.2 LaBrooy, R. and Dransfield, P., On the Time Response Optimisation of

Hydraulic Control Systems, Proc. 4th Intl. Fluid Power Symp., BHRA, 1975.
9.3 Young, M.R., Digital Simulation of Hydraulic Control Systems, M.Eng.Sc.

9.4 Meichke, J., An Introduction to Computer-Aided Design, Prentice Hall, 1968.
9.5 Sugie, N., An Extension of Fibonaccian Searching to Multidimensional Cases,

IEEE Trans. on Auto. Control, AC-9, 1964.
9.6 Box, G.E.P., Evolutionary Operation: A Method for Increasing Industrial

Productivity, App. Stat. V6, N2, 1957.
9.7 Nelder, J.A. and Mead, R., A Simplex Method for Function Minimisation,
Computer Jrl., V7, 1964-65.
195
9.8 Jacoby, S.L., Kowalik, J.S. and Pizzo, J.T., Iterative Methods of Non-
linear Optimisation Problems, Prentice Hall, 1972.
9.9 Box, M.J., A New Method for Constained Optimisation and a Comparison with
Other Methods, Computer Jrl., VS, 1965-66.
9.10 Guin, J.A., Modification of the Complex Method Of Constrained Optimisation,

Computer Jrl., VI0, 1967-68.
9.11 Dransfield, P. and LaBrooy, R., Hydraulic Systems with Precision Reflexes,
Machine Design, May 20, 1976. The example developed in this article came
from Fluid Power Control, Blackburn, Reethof, and Shearer.
CHAPTER 10
PHENOMENA WHICH CAN AFFECT SYSTEM RESPONSE
i0.1 INTRODUCTION
Phenomena which are inherent in working hydraulic control systems to the extent
that they are always present have been included in the modelling procedures dealt
with in this Text. Included in this inherent phenomena category are
internal leakages through running clearances,

viscous friction associated with components in relative motion,
various R , C , and I effects,
minimal coulomb friction.
There is another category of phenomena whose constituents may or may not be

present and which if present will affect adversely the response of a hydraulic control
system. These parasitic phenomena include
excessive coulomb friction,

cavitation,
hydraulic backlash,
valve flow forces,
hydraulic lock,
contaminated system fluid.
Some can be allowed for to some degree, if necessary, in system simulation models.
It is better that components and systems be designed and operated so that parasitic
phenomena do not appear. To achieve this requires an understanding of the phenomena
and an appreciation of how each can be avoided, or its affects on system response
minimized.
197
With the exception of coulomb friction, which has been considered already, each
of the mentioned parasitic phenomena will be discussed briefly. References for
further study are provided.
i0.2 CAVITATION
i0.2.1 General Discussion
A decision to use a hydraulic system for a particular control application is

likely to be based on needs for
fast and stable point-to-point positioning of large resisting loads, or

speed control of large rotary loads,
accurate and stiff location of positioned loads, or holding of rotary

speeds,
accurate control of force or torque.
To provide these characteristics, the fluid used in the system must retain a high bulk
modulus as it passes around the operating system.
Quiescent hydraulic system fluids at normal temperatures and pressures have a

high bulk modulus. That is, they are highly resistant to compression and have the
potential to impose fast, stiff, and accurate drives on large resisting loads.
Anything which reduces the high natural bulk modulus of a working hydraulic fluid,
even temporarily, reduces the system's performance potential.
Quiescent hydraulic fluids contain gases. Typically, a hydraulic mineral oil

at atmospheric temperature and pressure contains some 9% by volume of air. This air
is not in the form of bubbles. It is dissolved in and is an integral part of the
fluid. Oil in this normal state is said to be saturated. The amount of air which
is absorbed and dissolved naturally in the oil is dependent on both pressure and
temperature. At higher pressures more air will be absorbed and dissolved, and vice
versa. Henry's law states that the solubility of a gas in a liquid is directly
proportional to pressure. At higher temperatures less air remains dissolved in the
oil, and vice versa.
If the pressure of a quiescent oil is changed slowly, air is absorbed and dissolved
(for increasing pressure) or comes out of solution and is rejected (for decreasing
pressure), as the natural saturation state appropriate to the particular pressure is
sought. The mechanism of rejection of previously dissolved air as pressure is
lowered is for bubbles of air to form and to float through the surface of the fluid.
198
This takes time, particularly with viscous fluids. Air is vastly more compressible
than is oil, and consequently lowering of pressure has far more effect on the volume
of the air component of the fluid than on the oil component.
The presence of dissolved air up to the natural saturation level has little
effect on the bulk modulus of an oil. Thus, air which remains dissolved in the
fluid will not affect the response of a hydraulic control system. However, the
presence of bubbles of air in the oil will render the fluid more compressible, as any
increase in pressure will quickly compress the bubbles of air. In these circumstances,
system response can be adversely affected. Taken to the extreme, if a large quantity
of air bubbles exists in the oil, the fluid suffers a drastic reduction in its bulk
modulus.
The fluid in a working hydraulic control system experiences frequent, large, and
sudden changes of pressure, in both directions. Thus, the saturation level of
dissolved air will be changing rapidly and the actual level of dissolved air will be
trying to adjust itself. This does not matter much, so far as system response is
concerned, while pressures are increasing, as the fluid will be sub-saturated.
However, when pressure is decreased rapidly, such as between a reservoir and a pump's
suction port, or downstream of the orifice of a control valve, the fluid can be super-
saturated and air coming out of solution as bubbles is a likely consequence. When
the pressure drop is from one positive value to another positive value well above
atmospheric pressure, this condition should not be serious. Nor should it seriously
affect system performance. The bulk modulus will be transiently lowered but will
remain high. If however the low pressure approaches atmospheric pressure or is below
it, the volume of air coming out of solution can be large, the bulk modulus of the
expanding fluid can approach zero, and system performance can be grossly degraded.
The formation of bubbles in the fluid due to air or gases coming out of solution
due to reducing pressure is known as gaseou~ er~uitation. In effect, the bubbles are
cavities in the liquid. Such cavities are highly compressible and cause reduction
of the fluid's bulk modulus, Once formed, gas cavities don't inu~ediately dissolve
if system pressure is again increased. It takes time for re-dissolving t o occur.
Thus a reduction in bulk modulus arising from gaseous cavitation can p e r s i s t
temporarily even if system pressure rises.
Quite apart from the natural gaseous cavitation described, a rapidly moving fluid
undergoing pressure changes, directional changes, and temperature changes in the
presence of air or other gases can entrain bubbles of the gas. This is likely to
occur for example if return lines to a reservoir are not drowned, or if there are air
leaks, n o matter how minor, in uncharged pump suction lines. Entrained bubbles
affect the fluid just as do bubbles of previously dissolved air. Thus the term
199
gaseous cauitGt~on can be regarded as including entrained undissolved gases.
Air entrained at low pressure can dissolve into the oil when pressure is increased,
raising its air content level towards the saturation state. This is highly
undesirable as the potential for gaseous cavitation, when pressure is inevitably
lowered at a later part of the hydraulic circuit, is increased. The oil in a well-
designed, well-operated, well-maintained hydraulic control system will tend to become
sub-saturated. Any air entrainment will negate this desirable condition.
Like any liquids, hydraulic system fluids will boil (vapourize) if either the
temperature is high enough or the pressure is low enough. Vapourization is
characterized by the development in the liquid of vaporous bubbles which in effect are
cavities in the liquid. Each liquid has a pressure, at which vapourization will
develop. If pressure is reduced to or below this vapour pressure, vapourization will
occur. A liquid's vapour pressure is temperature dependent, increasing as temperature
is increased. Thus, fresh water boils at 100C at sea-level atmospheric pressure
0.i MPa. That is, at 100C, its vapour pressure is 0.i MPa. At temperatures below
100C, the water's vapour pressure is below 0.i MPa and if atmospheric pressure is
maintained water re/nains in its liquid state. Water at 21C will vapourize if its
pressure is reduced to 0.0024 MPa, a strong vacuum state. Hydraulic fluids exhibit
similar characteristics. Mineral oils have vapour pressure characteristics lower
than those of water and are consequently less susceptible to cavitation. Water-based
hydraulic fluids have vapour pressure characteristics more like water.
A fluid's susceptibility or its resistance to vaporous c~i%~t{on can be

described by its cavitation nunlber
P - Pv
Nc = y p ~2
where P is the static pressure present

Pv is the appropriate vapour pressure
p is the density of the fluid
is the local velocity of the fluid
Y is a coefficient for the particular fluid.
Low Nc indicates a high tendency to cavitate and vice versa. Hence high Nc ,
which results from low Pv , low X and low p , is desirable.
Summarizing, cavitation occurs if cavities develop in the working fluid. A cavity

indicates the presence of a bubble of air, gas or vapour. Cavities can develop due
to
200
air or gas previously dissolved in the oil coming out of solution

(gaseous cavitation)
air or gas being entrained in the oil (gaseous cavitation)
pressure of the oil falling below its vapour pressure, whereon the oil
boils (vaporous cavitation).
In working hydraulic control systems, gaseous and vaporous cavitation can co-exist.
Gaseous cavities are likely to form as soon as system pressure is reduced below the
appropriate saturation value~ which can be a quite high pressure. Vapour cavities
will not form until pressure drops below the appropriate vapour pressure which is
normally around or below atmospheric pressure.
The presence of cavities in the working fluid reduces its bulk modulus. If the
reduction is dramatic, system response will be degraded.
The presence of dissolved air, and its coming and going as pressure changes, is a
natural phenomena. Little can be done about it. Unless accompanied by vaporous
cavitation, it is not usually of serious consequence. Entrained air is always un-
desirable and can seriously degrade system performance. Good reservoir design
and good suction line design can minimize the presence and effect of entrainment.
Vaporous cavitation can be avoided by keeping pressures at all parts of the system
above the fluid's vapour pressure. If necessary, reservoirs can be pressurized
and/or main pump suction lines charged.
Cavitation not only degrades system response and performance. It can also damage
the surfaces of components due to very high instantaneous pressures generated as
cavities implode. It can also damage the system fluid by oxidation due to the
very high local temperature generated as bubbles of air or gas or vapour are
compressed when pressure increases. Cavitation also degrades a system's
performance by generating offensive noise.
Refs. i0.I to 10.3 provide more detailed discussions of cavitation and its
relevance to hydraulic control systems.
i0~2.2 Cavitation in Modelling and Simulation
Fully-developed vaporous cavitation c~n cause the bulk modulus of the system
fluid in the region of cavitation to be zero. Under these conditions the pressure
in the cavitating region is effectively zero and no power will be being transmitted
from the region. The zero bulk modulus, zero pressure, zero power condition applies
only to the region of cavitation and is usually highly transient.
201
A dynamic simulation indicates the presence of vaporous cavitation if it shows a

pressure at or below atmospheric zero. Unless steps are taken to prevent it, a
simulation can happily show and proceed with a negative pressure. As negative
pressures (absolute) are unreal, a simulation becomes unrealistic if it allows the
development of negative pressures. The pressure of a cavitating liquid remains
effectively zero. However, it is not adequate in a simulation to simply hold a
negative pressure development at zero until it again becomes positive. Though this
appears to reflect the physical reality, it does not recognize the flow deficiencies
associated with the computed negative pressures. That is, there is no recognition
of cavity formation. This could lead to errors in load response predictions, though
the errors would probably be transient in most situations.
McCloy [10.4] provides an approach for allowing for cavity formation in simulation
models of hydraulic cylinders. He also suggests that the model complication required
is rarely justified on the basis that its affects on cylinder response are minimal.
Rogers [10.5] found it necessary to include modelling of cavity formation in simulation
studies of hydraulic backlash, to be discussed in Section 10.3.
The concensus of opinion appears to be that while cavitation in hydraulic control

systems remains an undesirable problem, its effects on load response are readily
eliminated or minimized by established system design procedures, and that it is very
rarely necessary to complicate dynamic simulation models by attempting to describe
cavity formation so long as these procedures are utilized.
10.3 HYDRAULIC BACKLASH
One of the most common applications for hydraulic control systems requires the
moving of an inertia load from one state (position or speed) to another in a
specified range. There is usually the requirement or expectation that the driven
load will be restrained accurately and stiffly in its new state prior to the next
instruction, which might be milliseconds, seconds, minutes, or hours later. For the
case of linear positioning, this objective is usually sought by using a hydraulic
actuator whose output is rigidly connected to the driven load and which receives oil
from a closed-centre or tandem-centre control valve. The intention is that when the
control valve is centred, both hydraulic lines joining the valve to the actuator will
be, and will remain, full of oil in a high bulk modulus state, thus stiffly locking
the load in its required position. With rotary drives, the intention is that both
the supply line and the discharge line of the rotary actuator be filled with oil of
a high bulk modulus and which is flowing at a constant rate.
202
Unfortunately, these states are not always achieved, and the driven load can
sometimes be easily moved about its desired state. This effect is called h y ~ l i c
backlash. Its basic characteristic is illustrated on Fig. i0.i. Ideally, any
attempt to move the located load should immediately induce a high resisting force or
torque from the hydraulic drive. If an atten~t to displace the load meets zero or
little resistance, at least over a band of displacement, hydraulic backlash exists.
In the circumstances described, hydraulic backlash can exist only if the hydraulic
fluid in one or both of the hydraulic lines of the actuator has a zero or very low
bulk modulus. It is assumed that the hydraulic components have normal resistance to
internal leakage and that the mechanical cc~ponents of the system and load are of
normal rigidity.
Not all hydraulic control systems are susceptible to the development of hydraulic
backlash. The phenomenon will develop in susceptible systems only under some
operating conditions, notably during rapid braking of inertia loads by valve or pump
displacement control. It tends to be transitory in nature and usually decays in
time after a load reaches equilibrium, though this time can be of significant duration.
Vaporous cavitation is pre-requisite for hydraulic backlash to develop, as a

deficiency of liquid oil in an actuator's line(s) implies the presence of vapour.
Hydraulic backlash is an effect or result possible if cavitation has occurred. Thus,
anti-cavitation additions to a hydraulic circuit, such as the check and relief system
illustrated on Fig. 10.2, act also to eliminate hydraulic backlash.
Rogers [10.5] made a substantial study of hydraulic backlash as it can develop

during valve-controlled braking of a system of the kind illustrated on Fig. 10.2.
With the cross-line relief system rendered ineffective, hydraulic backlash was quite
substantial, up to 3 radians, and could persist for up to 20 minutes after the load
was stopped. Rogers describes the mechanism of hydraulic backlash development. It
is attributable to flow deficiencies in the actuator lines, relative to load motion,
as the control valve is centred. Rogers' experiments and simulations show clearly
the existence of both cavitation and hydraulic backlash as the rotating inertia load
is slowed by braking. Rogers allowed the possibility of cavitation in his simulation
model. Ref. 10.6 provides a description of some of this work.
In some circumstances, hydraulic backlash might provide the explanation for an

observed response aberration or an aberration between an observed and a predicted
response.
203
Resisting force on torque

developed by the hydraulic system
!
/
!
, /
/
!
Displacement
applied to load
/
W
FIG. i0.i HYDRAULIC BACKLASH CHARACTERISTIC
rl
L.~l
tl-,,~
.J
It
FIG. 10.2 ANTI-CAVITATION CIRCUIT ADDITIONS

204
10.4 FLOW FORCES IN VALVES
In this Text, the relationships required for including valve characteristics in

the dynamic models of hydraulic control systems have been represented by R effects
which directly relate output flowrates tc input control signals. For example,
the manually controlled 4-way valve has been represented by systems of

equations centred around the relationship
Q = Kv - Xv P
pressure relief valves have been represented by
Qr = 0 if P ~ Pset
= Kr - (P - Pset) if P > Pset
electrohydraulic servovalves have been represented by systems of equations

similar to those for manually-operated valves.
The thrust of the Text has been not the accurate simulation of individual hydraulic
components, but rather the use of simplified representations of them which allow
reasonable predictions of the major system response characteristics such as response
of the driven load, and pressure developments in the system.
valves experiencing flow develop force reactions on their moving component due
to the flow. These parasitic flow forces are undesirable and can be minimized by
good design. As they may cause or account for unexpected system performance they
should be appreciated.
The origins and nature of flow forces can be appreciated by considering the
simple spool configuration of Fig. 10.3. In the centred state, the supply pressure
acts on the two internal spool land faces equally and the pressure forces on the spool
sum to zero. If the spool is moved to the right, as shown, the left control port
remains blocked while the right control port experiences flow. The left internal
spool land face continues to experience stagnant fluid at uniform pressure. However,
the fluid flowing at relatively high velocity through the right-hand port gains
kinetic energy, and this can only be achieved at the expense of pressure (potential
energy). Thus, in the region of the jet formed by the partly-opened orifice, the
pressure acting on the internal spool land face is reduced as shown. The axial
force due to pressure on the right-hand spool land is reduced below that on the
left-hand and a net axial force exists on the spool. The axial flow force acts to
close the valve. If not resisted, it will do so. With manually controlled valves,
the manual operator, possibly aided by a detente mechanism, usually can overcome the
flow force and hold the valve open. It may become a problem with large valves.
Flow forces axe more significant with electrohydraulic valves, as their axial
205
operating force is limited. The problem is worst with electrohydraulic servovalves,

because of their sensitivity and the accuracy of flow required of them. There are
spool-port design techniques which, though they add to expense, reduce flow force
magnitudes.
Most of the major texts on hydraulic control systems provide details (e.g.
Refs. 10.7, 10.8) allowing calculation of steady-state flow forces. Dynamic flow
forces are less readily assessed.
Flow forces in valves are rarely of significance in predictive simulations of

hydraulic control systems. Where necessary allowances can be made.
10.5 HYDRAULIC LOCK
Piston pumps and motors and spool-type valves rely only on the close fit of
piston to bore to minimize leakage past the piston due to axial pressure drop. The
fine clearances needed demand near-perfect geometry of both piston and bore, which is
a main reason why hydraulic piston and spool components are expensive to produce.
In operation, pistons and spools experience a number of forces, including
pressure area force in their axial direction,
inertia force due to their own mass,
viscous drag force, in their axial direction, due to viscous leakage flow
in the clearance between piston and bore, and due to viscous shear of the
oil film due to axial motion of the pistQn,
flow forces, as previously discussed,
mechanical friction due to any influence which causes the piston and bore
to be in contact instead of the piston "floating" in the clearance oil film.
The first four of these effects are inherent and are well understood. Single-
ended piston devices (pumps, motors, some pressure control valves) experience large
but expected end pressure forces which are mechanically reacted in the device. Spool
pistons axe usually symmetrical so that the net axial pressure force is zero. These
devices have the potential to be operated by a miniscule axial force, which has only
to overcome the inertia and the viscous drag forces. A piston 1 cm diameter, 2 cm
long, with diametral clearance 0.01 ram, and with axial pressure drop of 20 MPa
experiences a drag force of only about 2 N . Valve flow can induce axial and
radial forces on valve spools, inhibiting their performance and response, as
206
Fx
i?
,x ...i,...
if/M////,////
[ ~
#
~, -
I
,
I
Fflow
Unequal pressure distributions

FIG. 10.3 ILLUSTRATING THE ORIGIN OF FLOW FORCE
a. Pump or Motor Configuration

~F_ _Bearing reaction force
/ / / / / /
/i//-
b. Tilting due to Unsquare or Non-Concentric Spring Support
I / / I / / / / ///
////////
C. Bent Spool
Lateral force
FIG. 10.4 SOME CAUSES OF UNSYM-M~TRICAL CLEARANCES
IN WORKING PISTON COMPONENTS
207
separately discussed. Such flow forces can be over i00 N .
Mechanical friction requires that a lateral force exists to force the piston into
contact with its bore. Then, any attempt to move the piston axially must overcome
the resulting friction, and the force required to induce axial motion would include a
component
Ffx = p Fz
where is coefficient of friction between piston and bore

Fz is the net radial force producing contact between piston and bore
x denotes axial direction
z denotes lateral direction.
with piston pumps and motors, the eccentric nature of the drive acts to induce
piston-bore contact as the piston is forced against the bore (Fig. 10.4a). With
pumps, the high speed reversing motion of the piston allows little time for lateral
and tilting displacement of the piston to fully develop, and also encourages good
hydrodynamic lubrication to be sustained. It takes time to squeeze out a film of
mineral oil. In any case, the large power input to a p u m p completely overwhelms its
piston friction effects, and piston friction is reduced to a negligible effect as far
as dynamic response of the driven load is concerned. This is not so with hydraulic
motors unless they too are run at reasonably high and constant speed. The well-
known tendency of hydraulic piston motors to stick or to be jerky at stall or slow
speeds is at least partly due to friction deriving from lateral piston displacement
and tilt arising from the eccentricity of the drive.
Most spool type control valves are intended to be in hydrostatic balance,

requiring minimal forces to move the spool in its bore. Yet, a spool which is free
in its bore when no pressure exists in it can become hard to move under pressure.
Quite apart from the flow forces, two other possible causes for this kind of valve
sticking are contamination lock and hydraulic lock. Contamination lock will be
discussed in Section 10.6. The remainder of the p r e s e n t section will introduce that
parasitic phenomenon, hydraulic lock.
Fig. 10.Sa shows a piston lying centrally in its bore with a high pressure Pl
at one end and a lower pressure P2 at the other end. Ideally, one would imagine
that the pressure distribution in the clearance is linear as shown, and that it is
the same along all axial generators on the piston surface. That is, the axial
pressure distribution at 8 = 0 is the same as that at 8 = w and at all other
values of @ , and the radial forces generated on the piston surface sum to zero,
leaving the piston floating freely in its surrounding oil film.
208
a Parallel Concentric b. Parallel, Eccentric
o
e I
/ / Z z ] ~' f / W / W / / / / / /
P2 P1
Fz = 0 FZ = 0
f / / / / / i /
C. Partial Tilt d. Full Tilt
/ / / / / / / ~ / / / / / / / / / /
/ / / / / / / f / / I / / / / / / / / /
e. Convergent, Taper, Tilted f. Divergent, Taper, Tilted
// / / ~" / / // /
// W z" / // / / /
/ - / / / / / / / /," / / / / / / / / / /
FIG. 10.5 CLEARANCE S~APES AND PRESSURE DISTRIBUTIONS

FOR VARIOUS PISTON-BORE CONFIGURATIONS
209
But what if the axial pressure distribution is n o t the same at different v a l u e s

of 8 ? For example, Fig. i0.5c illustrates a humped pressure distribution at
@ = 0 , and a sagging pressure distribution at 8 = w . Both still start, as they
must, from P1 and end at P2 . The area under a pressure distribution represents
a radial force generated on the piston. Thus, the 8 = 0 distribution exerts a
downward force on the piston larger than the upward force of the 8 = ~ distribution,
and the piston is forced downwards towards contact with the bore. Frictional
resistance against axial motiOn can then be expected.
Friction between piston and bore arising frcm uneven axial pressure distribution
around a piston is known as hydraulic lock.
Under what conditions can unequal axial pressure distributions ~ radian apart
arise? A n d under what conditions will the n e t lateral force of such pressure
distributions cause decentring of the piston towards contact with the bore? These
questions have been studied fairly extensively, for example in Refs. 10.9, 10.10,
I0.II. If for any reason, the actual axial clearance between a piston and its bore
is diverging in the direction of axial pressure drop, hydraulic lock can develop.
The attitude of a piston in its bore can rarely be guaranteed. Even in the
ideal case of Fig. 10.5a, why should the piston be concentric with its bore? Its
own mass can cause it to sink into the clearance oil film. The piston's supporting
mechanism m a y cause it to tilt (Fig. 10.4a and b). A slight banana bend in a long
spool can cause the situation of Fig. i0.4c. A geometrically perfect piston might
have any of the attitudes shown in Fig. 10.5a to d. Also, geometric perfection of
piston and bore (soundness, parallelism, straightness) is impossible to achieve.
Imperfections contribute to uneven clearances and possible uneven axial pressure
distributions. Lack of parallelism, in either or both components, can generate the
clearance situations illustrated on Fig. 10.5e and f.
Fig. 10.5 includes the pressure distributions likely to be associated with various
piston-bore clearance ccnfigurations. Each indicates the net lateral force Fz which
will be generated by the pair of diametrically opposite pressure distributions shown.
For the parallel concentric and parallel eccentric cases (a and b) Fz is zero. For
the tilted piston cases (c and d) Fz acts to decentre the piston towards contact
with the bore. With the convergently-tapered clearance situation (e) Fz is finite
but acts to centre the piston in its bore, a highly desirable situation w h i c h holds
whether the piston is tilted or parallel in its bore. with the divergently-tapered
clearance (f) Fz acts to decentre the piston whether or not the piston is tilted.
The decentring lateral force necessary for hydraulic lock to develop will be present
if there is a diverging axial clearance due to piston tilt or component tapering.
210
Improved manufacturing techniques and quality control reduce the incidence of

hydraulic lock. Grooving of piston lands reduces the magnitude of hydraulic lock
by ensuring equal pressure around the piston at each groove. The expansion of the
bore and compression of the piston due to pressure at their high pressure end acts to
produce the desirable convergently-tapering clearance situation. Care should be
taken that spools and bores are not bent ~xially, as this can result in the tilted
piston configuration.
Hydraulic lock, if it develops, can be slight or substantial. If it develops

in pumps, relief valves, or manually-controlled valves, it is usually overcome by
brute force. In valves with limited operating forces, such as electrically
controlled valves, it can cause complete jamming and malfunction. Such valves
sometimes use electrical dither of the spool to keep the spool in motion. This
prevents static friction and acts to inhibit the development of hydraulic lock,
which is a time-dependent phenomenon at its worst if susceptible components are left
stationary under pressure.
It would be most difficult to allow for possible hydraulic lock in dynamic

simulation models, although a constant Fz force could be calculated and included
in particular circumstances. It is far better to take steps to ensure that hydraulic
lock is not a problem.
10.6 CONTAMINATED FLUID
Physical contamination is always present in hydraulic system fluids. Oil newly

supplied contains millions of particles per litre. A system with appropriate
filtration will clean up and maintain its fluid to a level at which the system
components will operate effectively and with acceptable life. Much work has been
done on studying the nature of contamination, the levels and distributions present in
working systems, the measurement and monitoring of contamination levels, the effects
on performance of contaminated fluids, and the tolerance levels of components to
contamination levels (Refs. 10.12, 10.13, 10.14).
Physically cc~%taminated fluid can erode and wear ccrnponents, degrading the
characteristics of control orifices and increasing internal leakage. It can also
cause direct physical malfunction of sensitive parts. For example, a sensitive
electrical control valve can silt up and jam if left stationary under pressure in
the presence of excessively cc~taminated oil.
Specification of a contamination level acceptable for a particular system is a

complex matter. It is insufficient to merely specify a gravimetric l~vel, i.e. so
many milligrams of contaminant per litre. The sizes, size distributions, and the
211
materials of the particulate are also important. So too are the materials and
clearances of the operating system components.
Fig. 10.6 illustrates crudely some mechanisms which could lead to performance
aberrations in piston components. Hard particles of sizes of the order of the
diametral clearance, and of half the diametral clearance, would be highly undesirable
as they could enter the clearance and lodge in it, especially if pressure was reduced
after entry. Silting at the entry to a fine clearance can affect pressure distribution
in the clearance and so give rise to hydraulic lock in sensitive components.
a. Particle of Size Half c. Tilting Due to Lodged P a r t i c l e
Diameter Clearance
i Ps
b. P a r t i c l e of Size D i a m e t e r
Clearance
FIG. 10.6 SOME MECHANISMS FOR CONTAMINATION LOCK
The person concerned with the dynamic response and performance of hydraulic
control systems should always be on the alert for aberrations in system performance
which may be attributable to the effects of fluid-borne contamination. Such
contamination will always be present. The idea is to keep it to levels at which the
system performance will not be affected, and which will allow long and reliable life.
212
10.7 CONCLUSION
Some phenomena which might or might not manifest themselves in working hydraulic
control systems have been discussed at an introductory level. There are other
parasitic phenomena. Thermal lock is one. Thermal lock occurs if working
components expand under the influence of temperature such that running clearances
become impaired. If manifested, parasitic phenomena will affect system performance
and response in a degradatory way. Most are not readily allowed for in dynamic
system models. If a system's actual response does not correlate well with a predicted
desired response, it could be due to one or more parasitic phenomena. For this
reason, the designer-analyst of hydraulic control systems should have familiarity
with them.
The phenomena discussed have been related mostly to piston configurations. They
may apply also to non-piston components.
10.8 REFERENCES
i0 .i MeCloy, D., Some Causes and Effects of Cavitation in 0il Hydraulic

Equipment, Proc. Conf. Fluid Power 1970, Fluid Power Soc., Melbourne,
1970.
10.2 Pearsall, I.S., Cavitation, Mills and Boon, London, 1972.
10.3 Wang, P.K.C. and Ma, J.T.S., Cavitation in Valve-Controlled Hydraulic

Actuators, Trans. ASME, Jrl. App. Mech., V30, 1963.
10.4 McCloy, D., Cavity Formation in Valve Controlled Hydraulic Cylinders,

Proc. I.Mech.E., V184, Part i, N52, 1969-70.
10.5 Rogers, K.J., On the Origins and Effects of Hydraulic Backlash in Tandem
Centre Valve Controlled Hydraulic Systems, Ph.D. thesis, Monash
University, 1976.
10.6 Rogers, K.J. and Dransfield, P., A Study of Hydraulic Backlash by

Simulation, Proc. Conf. on Control Eng., Inst. Eng. Aust., Melbourne,
1979.
10.7 Merritt, H., Hydraulic Control Systems, Wiley, 1967.
10.8 McCloy, D. and Martin, H., Control of Fluid Power, Longman, 1973.
10.9 Blackburn, J.F., Contributions to Hydraulic Control, Part 5, Lateral

Forces on Hydraulic Pistons, Trans. ASME, V75, 1953.
10.10 Manhajm, J. and Sweeney, D.C., An Investigation of Hydraulic Lock,

Proc. I.Mech.E., V169, N42, 1955.
10.11 Dransfield, P., Bruce, D.M., and Wadsworth, M., A General Approach to
Hydraulic Lock, Proc. I.Mech.E., V182, Part i, N27, 1967-68.
10.12 Brown, J.F.C., Contamination Control in Hydraulic Systems, Proc. 4th

Intl. Fluid Power Symposium, BHRA, 1975.
213
10.13 Tessman, R.K. and Fitch, E.C., Contaminant Induced Wear Debris for
Fluid Power Components, Proc. I.Mech.E. Technology Conf., Swansea,
U.K., 1978.
i0.14 Fitch, E.C., An Encyclopedia of Fluid Contamination Control for

Hydraulic Systems, Fluid Power Research Center, Oklahoma State
University, U.S.A., 1979.
CHAPTER 11
CONCLUSION
It is hoped that the Text has provided the reader with a reasonably deep
appreciation of the potential and the use of bond graphs as a key link in improving
his (read his or her) ability to predict the dynamic response of hydraulic control
systems. I encourage the reader with the desire but not the confidence to effect
such analyses, to try the bond graph approach. The reader already confident of his
ability to perform such analyses on proposed systems on the drawing board, may well
see bond graphs as just another modelling technique. I challenge such readers to
dig in sufficiently to be able to properly try the bond graph approach to simulation.
At the very least you will find it enlightening.
The Text has as its philosophy
that conventional static design procedures allow proposal of a system which

should carry out the desired functions,
that it is insufficient merely that the required functions will be induced;

the quality of the system actions and the time frame during which they occnlr
are also important, often vitally so,
that predictive dynamic response analyses should be carried out as part of

the system design procedure,
that bond graphs provide a key link in the confident development of dynamic
models,
that digital simulation is the way to extract response information from

mode i s,
216
that procedures exist which allow reasonable optimization of system response,
that the various phenomena inherent to the response of hydraulic control

systems should be clearly appreciated,
that there are phenomena parasitic to the response of hydraulic systems and
that the system designer should be aware of these and of their possible
effects.
It is not claimed that the Text is exhaustive. There are bond graph techniques,
inherent and parasitic phenomena, optimization techniques, and related matters which
have been barely referred to or neglected. There are however sufficient material
and e x a ~ l e s to allow the hydraulic system designer-analyst to investigate many
systems, and References from which he can extend his knowledge when ready. The
Author claims no omnipotence in the area of dynamic performance of hydraulic control
systems - - simply a reasonable knowledge and experience of it which has been
significantly enhanced by his adoption of bond graphs.
APPENDIX 1
STATIC DESIGN APPROACH
INTRCDUCTION
A considerable number of books provide information and techniques for the design
of hydraulic control systems. Some are included in the References of this Appendix.
The designer rarely needs to design a major component. His task is to conceive a
strategy appropriate to the particular application, to represent this strategy on a
circuit diagram, and to select the components for his system from a wide range of
commercial stock most of which will be available on-the-shelf. There may be delays
in supply if the shelf happens to be in another country~
It has been assumed that the readers of this Text are reasonably familiar with
hydraulic control systems and to some degree with their design. This Appendix will
only outline a basic approach for arriving at a hydraulic system capable of
performing a specific task.
POINTS TO BE CONSIDERED
Appreciation of the task to be performed

type, size, mass, shape of load(s) to be driven;
motions required; type, size, speed, frequency of changes and reversals;
location of the task; factory, field, airborne, dockside, mobile, etc.;
working environment; dust or dirt levels, humidity, temperatures, noise levels,
space limitations.
Mode of operation and control

manual or automatic control;
218
continuous or intermittent operation;

how much full-power operation.
Safety requirements
protection of operators and others;
protection of the system;
adjacent machinery or equipment.
Reliability
life of system;
likely maintenance requirements;
access to systems;
ease of maintenance;
availability of maintenance expertise and facilities;
operator ability.
Cost of the system

capital cost;
operating cost;
maintenance cost.
Other factors
will operating noise be a problem?
does the client have any personal whims or requirements? Layout, preferred
maker of components, compatibility with existing system, etc.;
Size, space, mass, or layout limitations.
AN APPROACH
I. Establish the functions, motions, and cycles required of the system.
2. Propose a hydraulic circuit diagram for the system showing only major components
(actuators, control valves, pumps). This diagram constitutes a design strategy.
3. Appreciate any basic constraints placed on the system, e.g. maximum system
pressure, pump drive speed.
4. Specify the maximum strokes required of linear actuators and/or the maximum
speed required of hydraulic motors.
5. Assess the mauximum force or torque required of each actuator, allowing for
inertias, friction, and external forces or torques the drive will be expected
to overcome.
219
6. Determine approximate actuator size (s) on the bases:
actuator piston area required = maximum force required/maximum pressure of

system
volumetric displacement per radian of motor = maximum torque required/

maximum pressure of system
7. Determine the maximum input flow requirement of each actuator on one of

following bases:
inflow rate required = axea of piston maximum piston velocity required
inflow rate required = volumetric displacement per radian maximum angular

velocity required
8. Determine the approximate requirements of the main control valve(s), on the

base s :
maximum flowrate required at the actuator(s) serviced by each valve;
maximum system pressure to be allowed;
type of control required, e.g. manual, electrical, closed-centre, etc.
9. Determine the approximate pump discharge rate required on the bases of:
the number of actuators to be driven simultaneously;
the maximum flowrate required at the actuators;
the maximum flow combination;
allowance for internal leakages of actuator(s), motor(s), and control

valve (s) .
i0. Select commercial actuator(s) and control valve (s). Refine approximate pump
discharge rate requirement accordingly.
ii. Select a commercial pun~p, considering
speed at which p ~ p is to be driven;
maximum pump discharge rate required;
maximum pressure to be allowed in the system;
volumetric efficiency of pump;
type of system and type of control (closed, open, tandem-centre, etc.);
choice between fixed displacement, variable displacement, over-centre

variable displacement, pressure cempensation, automatic power limiting and
control features, etc. ;
possible use of an accumulator.

220
12. Decide on oil filtering strategy. Select filters.
13. Decide if any intermediate valving or control circuits are required and select
suitable components.
14. Decide hydraulic line or hose sizes (length, internal diameter, strength).
15. Update the circuit diagrams.
16. Estimate the maximum temperature rise in the system oil due to working of the
system.
17. Decide on reservoir size and if oil cooling is required.
18. Select the primary drive for the pump.
19. Allowing for pressure drops, internal leakages, and for the actual sizes and
ratings of the commercial items, refine all calculations.
20. Check that the system meets the action and performance requirements.
21. Estimate the cost of system components and of building the system.
22. Optimize the system and component selection on the bases of:
minimum cost;
adequacy of performance;
minimum number of components.
23. Finalize the design, component selection, and the estimate of cost.
The procedure outlined requires use of static or equilibrium relationships and

component manufacturers' data. There can be no certainty as to the actual dynamic
response and performance of the proposed system. Dynamic analysis of the type which
is the subject of this Text should begin after Step ii.
REFERENCES ON BASIC DESIGN
Guillon, M., Hydraulic Servo Systems: Analysis and Design, Butterworths,

1969.
Yeaple, F.D. (Ed.) , Hydraulic and Pneumatic Power and Control, McGraw-Hill, 1966.
Merritt, H., Hydraulic Control Systems, Wiley, 1967.
Kibble, J.D., Basic Factors in the Design of Hydraulic Circuits, Proc. I.Mech.E.,
V.184, Part I, N.I, 1969-70.
Prokes, J., Hydraulic Mechanisms in Automation, Elsevier, 1977.

Khaimovich, E.M., Hydraulic Control of Machine Tools, Pergamon, 1965.
APPENDIX 2
SI CONVERSION FACTORS
T h e U S A is t h e m a j o r i n d u s t r i a l n a t i o n n o t y e t f u l l y u s i n g the SI s y s t e m . The
f o l l o w i n g t a b l e a l l o w s c c ~ v e r s i o n o f u n i t s co~muonly u s e d in t h a t c o u n t r y ' s f l u i d
p o w e r i n d u s t r i e s to b e c o n v e r t e d to SI.
Basic Common
Quantity Basic Conversion Alternatives
SI U n i t USA Unit
Basic Units
Length metre (m) inch (in) in/39.37 = m in 25.4 = mm

in 2.54 = cm
ft/3.28 = m
Mass k i l o g r a m (kg) pnd. m a s s (ibm) i b m 0 . 4 5 3 6 = k g slug 14.6 = kg
Time s e c o n d (s) second 1
Current a m p e r e (A) ampere 1
Temp.(gauge) deg.cel. (C) deg. fah. (OF) (=F - 3 2 ) / 1 . 8 = C
Temp. (absol.) deg.kel. (K) d e g .Rank. (OR) R / I . 8 = K
Derived Units
Area m2 in 2 in2/1550 = m 2 in 2 , 6 . 4 5 2 = c~n2
Volume m3 US g a l l o n US gal/264 = m 3 US g a l 3 . 7 8 5 =
Density kg/m 3 Ibm/ft 3 lhm/ft3,16.02 = kg/m 3
F o r c e (also newton (N) ibf ibf 4.448 = N
weight)
Torque N -m i b f " in (Ibf-in)/8.85 = N m
Rot.inertia kg - m 2 i b m - in 2 ( i b m ' i n 2 ) / 3 4 1 3 = kg-m;
Pressure pascal (Pa) i b f / i n 2 (psi) psi - 6895 = Pa )si/145 = M P a
Pa = N/m 2 psi/14.5 = bar
Energy(also joule ft " ibf (ft'ibf) 1.356 = J
work) (N - m)
Power w a t t (W) (J/s) h / p o w e r (hp) hp 746 = W ]p/1.341 = kW

Elec.potent. v o l t s (E) volts 1
Elec.resist. ohm ohm 1
Linear vel. m/s ft/s (ft/s)/3.28 = m/s
L i n e a r acc. m/s 2 ft/s 2 (ft/s 2 ) / 3 , 2 8 = m / s 2
Flowrate m3/s USgal/min (USg/min)/15840 = m3/s USg/min-3.785 = /min
Angle rad tad 1 rev = 2~ tad
Ang.velocity rad/s rad/s 1 rpm 2~/60 = rad/s
Frequency h e r t z (Hz) Hz 1 c y c l e / s e c = Hz
Dynamic N s / m 2, o r oentipoise cP/E3 = Pa s
viscosity Pa - s (cP)
Kinematic m2/s centistoke cSt/E6 = m2/s
viscosity (cst)
Note: i0 n is s h o w n as En
i 0 -n is s h o w n as E-n
APPENDIX 3
DYNAMIC RESPONSEmA BIBLIOGRAPHY
There is a remarkably large number of English language papers concerned with the
dynamic response and performance of hydraulic control systems. They appear in a wide
variety of journals, with fairly regular contributions in the Proceedings of the
Institution of Mechanical Engineers, the Proceedings of BHRA's International Fluid
Power Symposia, the Bulletin of the Japan Society for Mechanical Engineers, and in a
number of the ASME journals. The present Bibliography is quite limited, choices for
it being based on whether a papex is directly referenced in the Text, whether it is
available in accessible literature, whether it is clearly relevant to the work of the
Text, and an attempt to provide the names of persons working in the area. Some
excellent older references are not included as they are referred to in later papers.
Books
Blackburn, J.F., Reethof, G., and Shearer, J.L., Fluid Power Control, M.I.T. Press,
1960.
Merritt, H.E., Hydraulic Control Systems, Wiley, 1967.
Guillon, M., Hydraulic Servo Systems: Analysis and Design, Butterworths, 1969
(translation from French).
McCloy, D. and Martin, H.R., The Control of Fluid Power, Longman, 1973.
Proke~, J., Hydraulic Mechanics in Automation, Elsevier, 1977.
Viersma, T.J., Analysis Synthesis and Design of Hydraulic Servosystems and

Pipelines, Elsevier, 1980.
Papers, Theses, Articles
Rausch, R.G., The Analysis of Valve Controlled Hydraulic Servomechanisms, Bell

Syst. Tech. Jrl., V38, 1959.
223
Wang, P.K.C., Mathematical Models for the Time Domain Design of Electrohydraulic
Servomechanisms, A.I.E.E. Trans. Applic. and Industry, VS0, 1961.
Wang, P.K.C. and Ma, J.T.S., Cavitation in Valve Controlled Hydraulic Actuators,
Trans. A.S.M.E. Jrl. App. Mech., V30, 1963.
McCloy, D. and Martin, H.R., Some Effects of Cavitation and Flow Forces in the
Eleetrohydraulic Servomechanism, Proc. I.Mech.E., V178, Pt.l, 1963-64.
Davies, R.M. and Lambert, T.H., Transient Response of a Hydraulic Servomechanism

Flexibly Connected to an Inertia Load, J. Mech. Eng. Sci., V6, 1964.
Dransfield, P. and Bruce, D.M., Leakage Flow Rate Past Pistons of A Hydraulic
System Components, J. of Aircraft, A.I.A.A., V5, N2, 1968.
Healey, A.I. and Stringer, J.D, Dynamic Characteristics of an Oil Hydraulic

Constant Speed Drive, Proc. I.Mech.E., V183, 1968-69.
Montgomery, J. and Lichtaravicz, A., Asymmetrical Lap in Hydraulic Servomechanisms,

Proc. I.Mech.E., V183, Pt.l, 1968-69.
Lepp, R.M., A Study of the Time Optimized Hydraulic Servomechanism Operating Under
Cavitation Conditions, Ph.D. Thesis, U. of Saskatchewan, Canada, 1968.
Takenaka, T. and Urata, E., The Dynamic Characteristics of oil Hydraulic Control
Valves, Bull. J.S.M.E., VI2, 1969.
Burrows, C.R. and Webb, C.R., The Design of Fluid Power Systems Using Analogue
Computer, Trans. Inst. Meas. & Control, V2, 1969.
Haines, D.F. and Davies, R.M., Harmonic and Transient Behaviour of an Electro-
hydraulic Servomechanism Test Rig, Control, VI3, 1969.
Nikiforuk, P.N. and Tsai, S., Detailed Analysis of a Two-Stage Four-Way Electro-
hydraulic Control Valve, J. Mech. Eng. Sci., VII, N2, 1969.
Urata, E., The Ramp Response of a Loaded Hydraulic Servomechanism, Bull. J.S.M.E.,
VI3, 1970.
Green, W.L. and Crossley, T.R., An Analysis of the Control Mechanism Used in
Variable Delivery Hydraulic Pumps, Proc. I.Mech.E., V185, 1970-71.
Schl~sser, W.M.J., A Contribution to the Analysis of the Characteristics of

Hydraulic Valves, Hyd. and Pneu. Power, V24, N7, 1971.
Ulrich, H.J., Some Factors Influencing the Natural Frequency of Linear Hydraulic
Actuators, Intl. of Mach. Tool Design and Research, VII, 1971.
Armstrong, P.J. and McCloy, D., Optimization of a Hydraulic Servo Using On-Off
Controllers, Proc. 2nd Intl. Fluid Power Symp., BHRA, 1971.
Alpay, S.A., The Influence of Dynamic Effects in Discharge Coefficients in Spool

Valves, Proc. 2nd Intl. Fluid Power Symp., BHRA, 1971.
Knight, G.C., McCallion, H., and Dudley, B.R., Connection Capacitance Effects in
Hydrostatic Transmission Systems and Their Prediction by Mathematical Model, Proc.
I.Mech.E., V86, 1972.
Bowns, D.E. and Worton-Griffiths, J., The Dynamic Characteristics of a Hydrostatic

Transmission System, Proc. I.Mech.E., V186, 1972.
224
Keating, T and Martin, H.R., Mathematical Models for the Design of Hydraulic
Actuators, Trans. Inst. Soc. of Am., 1972.
Smith, C.K., Sebesta, H.R., and Bose, J.E., Stabilization of a Hydromechanical

Steering System, SAE Paper 720791, 1972.
Unruh, D., A Standard Format for Mathematical Models of Fluid Power Systems, Proc.
26th N.C.F.P., 1972.
Armstrong, P.J. and McCloy, D., Design of High Speed Hydraulic Position Servo-
mechanisms Using Optimization Techniques, Proc. I.Mech.E., V187, 1973.
Young, M.Y., Digital Simulation of Hydraulic Control Systems, M.Eng.Sc Thesis,

Monash University, Australia, 1973.
Barnard, B.W., The Dynamic Response of a Hydraulic System, M.Eng.Sc. Thesis, Monash
University, Australia, 1973.
Chiappulini, R., A Review of Hydraulic Drives and S e r v o d r i v e s - - S o m e Actual

Problems, Proc. 3rd Intl. Fluid Power Symp., BHRA, 1973.
Martin, K.F., Design Approximations for Hydraulic Servos with Non-Symmetrical Oil
Volumes, Meast. & Control, V6, 1973.
Muto, T. and Hattori, T., A Study on the Unstable Phenomena in Hydraulic Driving
System and Direct Servosystem, Bull. J.S.M.E., VI7, 1974.
SchlSsser, W.M.J., A Contribution to the Study of Analogies of Power Transmissions

in Machines, Proc. I.Mech.E., V188, 1974.
Helgestad, B.O., Foster, K., and Bannister, F.K., Pressure Transients in an Axial
Piston Hydraulic Pump, Proc. I.Mech.E., V188, 1974.
Dreymuller, J., Pilot-Operated and Directly Actuated Pressure Control with variable
Delivery Axial Piston Pumps, Proc. 4th Intl. Fluid Power Symp., BHRA, 1975.
La Brooy, R. and Dransfield, P., On the Time Response Optimization of Hydraulic

Control Systems, Proc. 4th Intl. Fluid Power Symp., BHRA, 1975.
La Brooy, R., On the Optimization of Hydraulic Control System Response, M.Eng.Sc.

Dransfield, P. and La Brooy, R., Hydraulic Systems with Precision Reflexes, Machine
Design, 20 May 1976.
Parker, G.A. and Desjardins, Y.C., Identification of an Electrohydraulic Servo

Operating in the Underlap Region of the Valve, Proc. Conf. Fluids in Control and
Automation, BHRA, 1976.
Martin, D.J. and Burrows, C.R., The Dynamic Characteristics of an Electrohydraulic

Servovalve, Trans. A.S.M.E., J.D.SoM.C., Dec. 1976.
Chong, F.K., Effects of Different Pressure ReLief Valves on Dynamic Response of

Hydraulic Control Systems, M.Eng.Sc. Thesis, Monash University, Australia, 1977.
Barnard B. and Dransfield, P., Predicting Response of a Proposed Hydraulic Control

System, Using Bond Graphs, Trans A.S.M.E., J.D.S.M.C., V99, Series G, 1977.
Le Vert, FOR., Dynamic Analysis of a High-Speed Electrohydraulic Transient Rod Drive

System, A.S.M.E. Paper 77-WA/Fi CS-8, 1977.
225
Martin, K.F., Ong, C.S., Tee, B.L., and Little, E., Dynamic Analysis of an
Electrically-Operated Pressure Control Valve, Proc. 5th I.F.P.S., BHRA, 1978.
Mirakawa, S., The Effect of Flow Force on Frequency Characteristics of Hydraulic

Motor Driven by Spool Valve, Bull. J.S.M.E., V21, N154, 1978.
Svoboda, J., Analogue and Digital Modelling in the Design of a Hydraulic Vehicular
Transmission, Proc. I.Mech.E., V193, 1979 .
Dransfield, P. and Teo, M.K., Using Bond Graphs in Simulating an Electrohydraulic

System, J.Franklin Inst., V303, N3, Pergamon, 1979.
amaguchi, A. and Ishikawa, T., Characteristics of Displacement Control Mechanisms

in Axial Piston Pumps, Bull. J.S.M.E., V22, 1979.
Akashi, H., Nakagawa, T., and O~umi, T., Analysis of Vibration in Hydraulic Drive
System, Hull. J.S.M.E., V22, 1979.
Davies, D. and Stecki, J.S., Hydraulic System A n a l y s i s - - M o d e l l i n g of Fluid

Transmission Lines, BFPR Jrl. VI4, N3, Oklahoma State University, 1980.
Stecki, J.S. and Cuthbertson, S.C., The Effects of Changeover to Fire Resistant
Fluids on Hydraulic Performance, Fire Resistant Hydraulics Jrl., VI, N2, pp 167-173,
FPRC, Oklahoma State University, 1981.
Ramachandran, S. and Dransfield, P., Interaction in a Two-Channel Loaded Electro-

hydraulic Drive, Proc. 6th Intl. Fluid Power Syrup., BHRA, 1981.
Backe, W., DSH Program System for Digital Simulation of Hydraulic Systems, Proc. 6th
Intl. Fluid Power Syrup., BHRA, 1981.
Bowns, D.E., Tomlinson, S.P., and Bonson, L.A., Progress Towards a General Purpose
Hydraulic System Simulation Language, Proc. 6th Intl. Fluid Power Syrup., BHRA, 1981.

Hydraulics

Uploaded by

Document Informationclick to expand document informationHydrulic control systems

Document Informationclick to expand document information

Copyright:

Available Formats

Hydraulics

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Hydraulics

Uploaded by

Copyright:

Available Formats

Lecture Notes in

Hydraulic Control Systems -

ISBN 3-540-10890-4 Springer-Vertag Berlin Heidelberg NewYork

the quality of their dynamic response is important to their proper operation,

prediction of dynamic response capability is best obtained via the

the inability of many designers to efficiently effect predictive dynamic

the use of p o ~ bon~ ~aphs is a natural and superior approach to the

digital simulation provides the most appropriate method for extracting

P.D. March 1981

C~PTER2 NOTATION AND UNITS 13

CHAPTER 3 CONVENTIONAL MODELLING PROCEDURES 19

CHAPTER 4 POWER FLOW MODELLING 29

CHAPTER 5 POWER BOND GRAPHS 48

CHAPTER 6 SOLUTION OF POWER FLOW MODELS 94

CHAPTER 7 SELECTING EQUATIONS AND COEFFICIENTS i01

CHAPTER 8 APPLICATIONS OF BOND GRAPHS 153

CHAPTER 9 OPTIMIZING DYNAMIC RESPONSE 183

CHAPTER i0 PHENOMENA WHICH CAN AFFECT RESPONSE 196

APPENDIX 1 STATIC DESIGN APPROACHES 217

APPENDIX 2 SI CONVERSION FACTORS 221

APPENDIX 3 DYNAMIC RESPONSE - - A BIBLIOGRAPHY 222

Hydraulic control systems are used

where relatively large forces or torques are required (industrial presses;

where fast, stiff response of resisting loads is required (machine tool

where close control of response is required (aircraft control surfaces,

where manual control over substantial powered motions is required (mobile

as the final actuator sub-system in complex automatically controlled

a. A Linear Drive Hydraulic Control System

b. A Rotary Drive Hydraulic Control System

FIG. i.i TYPICALHYDRAULIC CONTROL SYSTEMS

Dynamic analysis can be

synthetical in nature, with the objective of ensuring good dynamic response

1.2 A DESIGN PHILOSOPHY

The stage Propose System requires the first corm~itment to a technology in

Control Valve Motion Desired Speed, rpm

Full Opez 5OO

Load Motion System Speed, rpm

FIG. 1.2 SOM~ COMMAND-AND-RESPONSE PAIRS

A decision is made then that the performance of the proposed system is

adequate (Response Acceptable; Finalize Design),

encouraging enough so that simple parameter adjustments will make it

relatively hopeless, requiring re-design of the proposed system (Response

Accepting that solution of mathematical models is relatively easy, the great

2. Define Required Response,

(Solve Model) " ~I 6. Compare I

FIG. 1.3 APPROACH FOR DESIGNING DYh~AM_IC RESPONSE

an adequate mathematical model of his proposed system.

1.3 DYNAMIC MODELLING

Conventional dynamic modelling procedures for control systems, such as

It appears that the conventional dynamic modelling procedures do not adequately

the need for easy-to-follow formality in the modelling process,

An engineering designer needs to be creative - - to have imagination ........for

Hydraulic control systems have some distinctive features relating to dynamic

They are usually of relatively small scale. In terms of a transfer function,

These latter two factors cause linear analysis techniques to be of very

They are power transducing, power transmitting systems. Typically, power is

Conventional modelling procedures do not particularly recognize these factors, nor

b. Power Bond Graph