0 Голоса «за»0 Голоса «против»

Просмотров: 13228 стр.Hydraulic Engine Mount Modeling, Parameter Identification and Experimental Validation

Dec 08, 2016

© © All Rights Reserved

PDF, TXT или читайте онлайн в Scribd

Hydraulic Engine Mount Modeling, Parameter Identification and Experimental Validation

© All Rights Reserved

Просмотров: 13

Hydraulic Engine Mount Modeling, Parameter Identification and Experimental Validation

© All Rights Reserved

- Hidden Figures: The American Dream and the Untold Story of the Black Women Mathematicians Who Helped Win the Space Race
- Hidden Figures Young Readers' Edition
- The Law of Explosive Growth: Lesson 20 from The 21 Irrefutable Laws of Leadership
- The E-Myth Revisited: Why Most Small Businesses Don't Work and
- The Wright Brothers
- The Power of Discipline: 7 Ways it Can Change Your Life
- The Other Einstein: A Novel
- The Kiss Quotient: A Novel
- State of Fear
- State of Fear
- The 10X Rule: The Only Difference Between Success and Failure
- Being Wrong: Adventures in the Margin of Error
- Algorithms to Live By: The Computer Science of Human Decisions
- The Black Swan
- Prince Caspian
- The Art of Thinking Clearly
- A Mind for Numbers: How to Excel at Math and Science Even If You Flunked Algebra
- The Last Battle
- The 6th Extinction
- HBR's 10 Must Reads on Strategy (including featured article "What Is Strategy?" by Michael E. Porter)

Вы находитесь на странице: 1из 228

Experimental Validation

by

Aaron A. Geisberger

A thesis

presented to the University of Waterloo

in fulfillment of the

thesis requirement for the degree of

Master of Applied Science

in

Mechanical Engineering

c Aaron A. Geisberger, 2000

I authorize the University of Waterloo to lend this thesis to other institutions or individuals for

the purpose of scholarly research.

Aaron A. Geisberger

I authorize the University of Waterloo to reproduce this thesis by photocopying or other means,

in total or in part, at the request of other institutions or individuals for the purpose of scholarly

research.

Aaron A. Geisberger

ii

The University of Waterloo requires the signatures of all persons using or photocopying this

thesis. Please sign below, and give address and date.

iii

Abstract

This thesis presents a modeling study of hydraulic engine mounts currently used in the

automotive industry. Nonlinear model aspects are developed and used with experimentally identified parameters to validate the model response characteristics. It is felt that this contribution

will help engineers in reducing mount design time, by providing insight into the effects of various

parameters within the mount.

In the thesis, various components and flow passages are assigned lumped parameters, in

order to arrive at a physical representation of the mount system. Bond graph techniques are used

to derive the differential system equations, which are used to develop the mechanical analogue

of the system. Of particular interest are the dynamic behaviors of the decoupler, inertia track

and bell assemblies. The nonlinear responses of these systems are considered, and in some cases

produce interesting behaviors that are typical of nonlinear systems.

An extensive set of experiments are conducted using a unique test apparatus to provide

numerical data for the various parameter identification techniques. These data also establish the

relative importance of several damping, inertia and stiffness terms. In addition, the measured

responses of the mounts to loading at various frequencies and amplitudes are compared to the

predictions of the mathematical model. The comparisons generally show a very good agreement

(better than 10%), which indicate that the analysis has been effective.

It is shown that the physical modeling in this thesis provides the appropriate system response over the full range of loading conditions (frequency and amplitude) encountered in practice. These results present an improvement on existing nonlinear models, which were limited

either to low frequency, high amplitude conditions, or use a piecewise linear approach. The results show the importance of several individual design parameters on the performance of the

mounts. Some areas requiring additional work, such as the fluid resistance terms for the inertia

track, and column inertia data for the decoupler orifices, are identified.

iv

Acknowledgments

I would like to thank my supervisors Dr. M.F. Golnaraghi and Dr. A. Khajepour, for their

input and support over the duration of this project. Also, the effort from my readers, Dr. G. Heppler and Dr. R. Macdonald, to promptly read my thesis was greatly appreciated.

Thanks are due for the financial support from Cooper-Standard Automotive, the Ontario

Graduate Scholarships and the Natural Sciences and Engineering Research Council (NSERC).

The cooperation and enthusiasm of everyone in the NVH Control Systems group at CooperStandard Automotive has made this research possible. Particularly, I would like to thank Rob

Bender, Carl Ohrling, Rob Sehn, Bernie Rice and Rob Paladichuk for all their help. Enjoy the

read.

I would like to take this opportunity to thank my parents, for their comfort and support

through all my endeavors, and the values they have taught me. Thanks also to my sister and

brother for providing happiness and inspiration to my life.

This thesis would not have been possible without the encouragement of friends, and the

many festivities. To: Barry & Tara, Brent & Jen, Dag, Gord, Gus, Shelley, Taylor, Tom & Kelly

and Thanh, I am very fortunate to have you as friends, cheers. Also, to my roommates, Kyle,

Matt, Mike and Trev, you kept life interesting during this work, thanks.

Now, I am off to windsurf.

Table of Contents

1 Introduction

1.1.1 Road Excitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.1.2 Engine Eccentricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.2 The Hydraulic Engine Mount . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.2.1 Determining Mount Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

1.2.2 Mount Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.3 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.4 Objectives and Thesis Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

17

2.1.1 Parameter and Variable Assignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.1.2 Bond Graph Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.1.3 System Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.1.4 Developing Two Linear Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2.2 Mechanical System Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.3 Nonlinear Model Extension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

2.3.1 Upper Chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

vi

2.3.3 Inertia Track . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

2.3.4 Decoupler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

2.4 Modeling the Additional Bell System. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

2.4.1 Lumped Parameter Assignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

2.4.2 Bond Graph Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

2.4.3 System Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

2.5 Mechanical System Model with a Bell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

2.6 Nonlinear Enhancement of the Bell Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

2.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

57

3.2 Parameter Study for Linear Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

3.2.1 Compliance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

3.2.2 Inertia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

3.2.3 Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

3.2.4 Effective Pumping Area. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

3.2.5 Linear Stiffness and Damping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

3.3 Nonlinear Decoupler Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

3.3.1 Low Frequency Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

3.3.2 High Frequency Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

3.3.3 Influence of Nonlinear Decoupler Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

3.4 Linear Bell Model Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

3.5 Parameter Studies on a Mount with the Bell System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

3.5.1 Compliance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

vii

3.5.2 Inertia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

3.5.3 Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

3.5.4 Area Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

4 Experimental Design

93

4.2 Design Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

4.3 Final Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

4.4 Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

4.5 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

4.5.1 Vessel Chamber Compliance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

4.6 Apparatus Capabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

5 Parameter Identification

105

5.1.1 Stiffness and Damping Parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

5.1.2 Effective Pumping Area Parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

5.1.3 Volumetric Compliance Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

5.2 Lower Chamber Compliance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

5.3 Inertia Track Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

5.4 Decoupler Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

5.5 Bell Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

5.5.1 Fluid Column Inertia and Resistance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

5.5.2 Chamber Compliance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132

5.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134

viii

136

6.2 Model Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

6.2.1 Inertia Track . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

6.2.2 Decoupler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

6.2.3 Inertia Track and Decoupler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

6.2.4 Inertia Track and Bell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

6.2.5 Inertia Track, Decoupler and Bell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

6.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152

153

7.1.1 Inertia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

7.1.2 Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155

7.2 Decoupler. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156

7.2.1 Inertia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156

7.2.2 Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159

7.2.3 Switching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160

7.3 Bell Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162

7.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163

8.1

164

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169

ix

B Frequency Domain Equation Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178

C Detailed Drawings of Experimental Apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . 184

D The Extended Kalman Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203

List of Tables

Table 3.1:

Table 3.2:

Table 3.3:

Table 3.4:

Table 3.5:

parameter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

Table 3.6:

Table 3.7:

inertia parameter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

Table 3.8:

Frequency modifications resulting from changes in the bell fluid column inertia

parameter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

Table 3.9:

Table 3.10: Frequency modifications resulting from changes in the effective pumping area

parameter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

Table 3.11: Frequency modifications resulting from changes in bell area parameter. . . . . . . . . 90

Table 5.1:

Table 5.2:

Inertia track, decoupler and bell parameters identified for the Cooper-Standard

mount. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

Table 7.1:

Calculated versus measured inertia parameters for the inertia track. . . . . . . . . . . . 154

Table 7.2:

Table 7.3:

Effective inertia and fluid column height for a simple orifice. . . . . . . . . . . . . . . . . . 156

Table 7.4:

Calculated versus measured inertia for the complete decoupler system. . . . . . . . . 158

Table 7.5:

Table 7.6:

Table 7.7:

xi

List of Tables

Table 1.1:

Table 1.2:

xii

List of Figures

Figure 1.1:

Frequency domain relative displacement transmissibility. . . . . . . . . . . . . . . . . . . . . . . 3

Figure 1.2:

High frequency excitations: (a) SDOF forced input excitation; (b) transmitted

force frequency domain plot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Figure 1.3:

Figure 1.4:

Figure 1.5:

characteristics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Figure 1.6:

Figure 1.7:

Typical mount dynamic characteristics: (a) Low frequency, large amplitude; (b)

High frequency, small amplitude. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Figure 2.1:

Figure 2.2:

Figure 2.3:

Figure 2.4:

Transmitted force to the mount base: (a) decoupler closed; (b) free floating

decoupler. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

Figure 2.5:

Figure 2.6:

Figure 2.7:

Figure 2.8:

Capturing bulge damping in the upper chamber: (a) previous model; (b) addition

of a resistance term in the compliance.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

Figure 2.9:

Figure 2.10: The nonlinear decoupler resistance: (a) sign convention; (b) switching

function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

xiii

List of Figures

Figure 2.11: Additional resistance function indicating influence of parameter. . . . . . . . . . . . . . . 36

Figure 2.12: Illustration of the arctangent function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

Figure 2.13: Isolated decoupler model that includes cage geometry: (a) geometry parameters;

and (b) pressure gradient breakdown. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

Figure 2.14: The decoupler area: (a) sign convention; (b) switching function. . . . . . . . . . . . . . . 40

Figure 2.15: Decoupler leak flow: (a) system model and sign convention; (b) nonlinear

resistance function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Figure 2.16: Cross section of a hydraulic mount with a bell system. . . . . . . . . . . . . . . . . . . . . . . . 44

Figure 2.17: Lumped parameter fluid system model including the bell. . . . . . . . . . . . . . . . . . . . . 45

Figure 2.18: Bond graph model that includes the inertia track, decoupler and bell. . . . . . . . . . . 46

Figure 2.19: Mechanical lever at top of mount. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Figure 2.20: Mechanical model of the hydraulic mount with a bell. . . . . . . . . . . . . . . . . . . . . . . . . 50

Figure 2.21: Mechanical model indicating transmitted force points.. . . . . . . . . . . . . . . . . . . . . . . . 51

Figure 3.1:

Dynamic stiffness and phase response of the linear low frequency model. . . . . . 59

Figure 3.2:

Low frequency linear model response of: (a) volume passing through inertia

track; and (b) pressure in upper chamber. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

Figure 3.3:

Low frequency linear model response, showing the stiffness response and liquid

column motion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

Figure 3.4:

High frequency linear model response, showing the stiffness response and liquid

column motion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

Figure 3.5:

Figure 3.6:

Figure 3.7:

Figure 3.8:

Linear model response for changes in the effective pumping area parameter. . . . 67

Figure 3.9:

Figure 3.10: Time domain simulation of nonlinear decoupler model: (a) decoupler volume;

(b) inertia track volume; (c) upper chamber pressure. . . . . . . . . . . . . . . . . . . . . . . . . . 71

Figure 3.11: Low frequency simulation of nonlinear model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

Figure 3.12: Dynamic stiffness characteristics of the nonlinear model under high frequency

excitations.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

Figure 3.13: Decoupler motion in the nonlinear model under high frequency excitations. . . . 73

Figure 3.14: Decoupler motion indicating region on instability and dependence on sine sweep

conditions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

Figure 3.15: Dynamic stiffness and phase indicating impact of sine sweep conditions. . . . . . . 75

xiv

List of Figures

Figure 3.16: Simulation of decoupler volume indicating the response to changes in the

decoupler parameter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

Figure 3.17: Dynamic stiffness and phase characteristics from the simulated linear model that

includes the bell and decoupler. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

Figure 3.18: High frequency linear model response of: (a) decoupler fluid column motion;

and (b) bell fluid column motion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

Figure 3.19: High frequency linear model response of: (a) upper chamber pressure; and (b)

bell chamber pressure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

Figure 3.20: Dynamic stiffness and phase response to changes in the upper chamber

compliance parameter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

Figure 3.21: Dynamic stiffness and phase response to changes in the bell chamber compliance

parameter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

Figure 3.22: Dynamic stiffness and phase response to changes in the decoupler fluid column

inertia parameter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

Figure 3.23: Dynamic stiffness and phase response to changes in the bell fluid column inertia

parameter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

Figure 3.24: Dynamic stiffness and phase response to changes in the decoupler resistance

parameter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

Figure 3.25: Dynamic stiffness and phase response to changes in the mount area parameter. 89

Figure 3.26: Dynamic stiffness and phase response to changes in the effective pumping

area. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

Figure 3.27: Dynamic stiffness and phase response to changes in the bell area parameter. . . . 91

Figure 4.1:

Experimental Approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

Figure 4.2:

Figure 4.3:

Figure 4.4:

Figure 4.5:

Figure 4.6:

Figure 4.7:

chamber. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

Figure 4.8:

Figure 5.1:

Figure 5.2:

xv

List of Figures

Figure 5.3:

low frequency high amplitude; (b) high frequency low amplitude. . . . . . . . . . . . . 108

Figure 5.4:

Test configuration for measuring the effective area and volumetric compliance

parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

Figure 5.5:

Figure 5.6:

Figure 5.7:

Upper compliance parameters identified at 15 psi mean pressure using the test

apparatus: (a) volumetric compliance; and (b) bulge damping. . . . . . . . . . . . . . . . 113

Figure 5.8:

Dynamic stiffness and phase response of the special mount, measured using

the test conditions in Table 5.1: (a) low frequency; and (b) high frequency

response.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

Figure 5.9:

Upper compliance parameters identified using the system model and test

conditions in Table 5.1: (a) volumetric compliance; and (b) bulge damping. . . 114

Figure 5.10: Test configuration for determining the volumetric compliance of the lower

chamber. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

Figure 5.11: Static measurements of lower chamber volumetric expansion versus pressure. 115

Figure 5.12: Test configuration for isolating the inertia track paramters. . . . . . . . . . . . . . . . . . . . 117

Figure 5.13: Parameter trends with respect to inertia track flow, for: (a) inertia; and (b)

resistance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

Figure 5.14: Isolated inertia track excitation plots including: (a) time domain segment

of random flow input; and (b) power spectral density of complete flow

perturbation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

Figure 5.15: Time domain segment of measured pressure differential versus model output

using estimated parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

Figure 5.16: Test configuration for isolating the decoupler parameters.. . . . . . . . . . . . . . . . . . . . 122

Figure 5.17: Steady-state time domain data of actual decoupler volume and pressure

differential at 12 Hz, showing: (a) nonlinear cage contact; and (b) linear

uncoupled response. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

Figure 5.18: Parameter trends with respect to decoupler flow, for: (a) fluid column inertia;

and (b) resistance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124

Figure 5.19: Time segment of the decoupler input flow and pressure differential, indicating

locations where leak resistance is identified. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

Figure 5.20: Decoupler state and parameter identification: (a) pressure differential across

decoupler; (b) decoupler volume state variable; and (c) decoupler switching

parameter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

xvi

List of Figures

Figure 5.21: Close up of the decoupler state and parameter identification: (a) pressure

differential across decoupler; (b) decoupler volume state variable; and (c)

decoupler switching parameter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

Figure 5.22: Cross section of Cooper-Standard mount that includes a bell plate. . . . . . . . . . . . 131

Figure 5.23: Test configuration for isolating the bell parameters. . . . . . . . . . . . . . . . . . . . . . . . . . 131

Figure 5.24: Parameter trends with respect to bell flow, for: (a) inertia; and (b) resistance. . 133

Figure 5.25: Illustration of the quarter plate model used to approximate the chamber

compliance, showing: (a) boundary constraints and 3 MPa evenly distributed

pressure load; (b) plate deflection in linear elastic region. . . . . . . . . . . . . . . . . . . . . 134

Figure 6.1:

Actuator model used to provide excitations with zero initial conditions. . . . . . . 138

Figure 6.2:

data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142

Figure 6.3:

inertia track. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142

Figure 6.4:

decoupler. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

Figure 6.5:

decoupler and inertia track. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

Figure 6.6:

decoupler and inertia track. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

Figure 6.7:

bell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

Figure 6.8:

inertia track and bell system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

Figure 6.9:

inertia track and bell system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148

Figure 6.10: Low frequency simulation versus measured response, on a mount including an

inertia track, decoupler and bell system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

Figure 6.11: High frequency simulation versus measured response, on a mount including an

inertia track, decoupler and bell system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

Figure 6.12: Comparison of high frequency response with and without the bell plate, using

both measured and modeled response. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

Figure 7.1:

Figure 7.2:

Measured resistance parameters for each test case: (a) flow resistance; and (b)

velocity damping. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156

xvii

List of Figures

Figure 7.3:

Fluid inertia tests of the decoupler orifice with: (a) seven holes; and (b) nineteen

holes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

Figure 7.4:

Figure 7.5:

cases investigated. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160

Figure 7.6:

Figure 7.7:

xviii

Nomenclature

Ab

Ad

Ad_holes

Ad_f nc

Ai

Am

Ap

Apiston

bb

bd

bi , bi1

bi2

br

Cb

Crig

C1

xix

Nomenclature

C2

Fp

FT

FT

fn

fo

hf luid

hgap

Ib

Id

Ii

kb

kr

k1

k2

Mplate

mb

md

mi

xx

Nomenclature

Pb

Ps1

Ps2

P1

P2

Qactual

calculated flow through the holding plate in the test apparatus [mm3 /s]

Qb

b

Q

Qd

d

Q

Qexcit

Qi

QL

QT

Radd

Rb , Rb1

Rb2

Rcb

Rd , Rd1

Rd2

Rd3

Ri , Ri1

xxi

Nomenclature

Ri2

RL2

RL2_f nc

R1

Vactual

calculated volume of fluid though the holding plate in test apparatus [mm3 ]

Vactual

Vd

Vd_ max

Vexcit

Vi

VT

VT

wdr

wn

Xact

Xb

Xd

k

X

xxii

Nomenclature

XT

Ps1,s2

damping ratio

phase angle [ ]

18

xxiii

Chapter

Introduction

Automotive design encompasses many engineering disciplines and has captured the minds of

many individuals. Using advanced technologies, the vehicles of today have been designed to

represent a zone of comfort, handling, and driving pleasure.

tuning the noise and vibration qualities to achieve the expected feel of a vehicle.

Reductions in body mass have increased engine-to-body weight ratios, which has raised

the level of noise, vibration and harshness within vehicles. More sophisticated hydraulic mounts

have been developed to address these issues. However, increased pressures to reduce vehicle

development cost has forced engineers to develop these complex hydraulic mount systems in a

limited time frame. The purpose of this work is to assist in reducing hydraulic mount development time by developing an effective hydraulic engine mount model and providing a physical

explanation of internal dynamics.

1.1

Vehicle occupants receive undesirable vibrations through one of two possible excitation sources.

The first source, from engine eccentricity, typically contains frequencies in the range of 25 to

200 Hz with amplitudes generally less than 0.3 mm [20]. This frequency range corresponds to

engine speeds of 750 to 6000 RPM for four cylinder engines. The second source of excitation

originates from road inputs and engine torque during harsh accelerations. Road inconsistencies

cause disturbances to the vehicle frame via the suspension system, whereas fierce accelerations

cause excessive engine torque and motion at the mounts. Excitations of this nature are typically

under 30 Hz and have amplitudes greater than 0.3 mm [20]. Each of these sources can be examined separately to demonstrate the desired mount characteristics, starting with the low frequency

excitations from road inputs.

1.1.1

Road Excitations

Low frequency, large amplitude excitations, transmitted from the road through the suspension

into the vehicle frame, can be represented as base excitation to the engine mount system.

To

demonstrate the desired mount characteristics, the system is simplified to a Single Degree-ofFreedom (SDOF) model illustrated in Figure 1.1a. Parameter m represents the engine mass,

while k and c denote the engine mount stiffness and damping characteristics, respectively. With

the disturbances from the vehicle body y(t), the equation of motion for this system is,

k(y x) + c(y x)

= m

x

(1.1)

This equation is converted to the frequency domain by assuming steady-state sinusoidal input and

output. Since the mount characteristics should minimize the relative displacement between the

engine and body, the frequency equations are manipulated to illustrate the relative displacement

transmissibility, given by (X Y )/Y . The equation becomes,

X Y

w

m

dr

=

2

Y

wdr m + wdr cj + k

(1.2)

where wdr is the driving frequency. Equation (1.2) is simplified using the nondimensional

damping coefficient = c/(2 mk) and plotted versus the nondimensional frequency ratio

p

r = wdr /wn , where wn = k/m in Figure 1.1b.

x(t)

m

k

c

y(t)

Relative

Displacement Transmissibility Amp

X-Y

Y

z=0.1

5 desired

operating

range

z=0.3

z=0.6

0.5

1.5

2.5

3.5

4.5

(a)

(b)

Figure 1.1: Low frequency excitations: (a) SDOF base excitation representation; (b) Frequency

domain relative displacement transmissibility.

Figure 1.1b represents the relative displacement transmissibility frequency domain plot.

This figure illustrates the desired operating frequency range for minimum engine to frame displacement. The driving frequency of base excitations should be below the systems natural frequency. Increasing the mount stiffness will yield higher natural frequencies and keep the large

amplitude vibrations within the displacement isolation region. Furthermore, increased damping will reduce the resonant amplitude and improve the relative displacements. Keeping these

characteristics in mind, the high frequency excitations will be examined next.

1.1.2

Engine Eccentricity

High frequency, low amplitude excitations, originating from engine eccentricity, can be represented as a force input to the mass of a fixed base SDOF system illustrated in Figure 1.2a. Forcing

on the engine mass Fo , represents the effective force input due to eccentricity and FT captures

transmitted force to the base. The resulting equation of motion for this system is

Fo = m

x + cx + kx

(1.3)

FT = cx + kx.

(1.4)

FT

Fo

Using this relationship and the steady state solution in the frequency domain, the following

equation is developed.

FT

cwdr j + k

=

2

Fo

wdr m + cwdr j + k

(1.5)

As in the base excitation example, non-dimensional damping and frequency ratio are used

to plot the frequency response in Figure 1.2b. In this case, engine vibrations should operate

in the isolation region, above the natural frequency of the engine and mount system. Engine

mounts with low stiffness will reduce the rigid body natural frequency such that the operating

frequency of the engine eccentricities are within this region. Low damping characteristics will

further reduce the transmissibility.

The above analysis has determined the desirable characteristics of engine mounts. During low frequency, high amplitude vibrations, the ideal mount should exhibit large stiffness and

damping characteristics to reduce relative displacement transmissibility. Whereas for high frequency, low amplitude vibrations the ideal mount should have low stiffness and damping characteristics. These conflicting properties indicate that an ideal mount system has stiffness and

damping characteristics dependent on the amplitude of excitation. These amplitude dependent

characteristics are achieved in hydraulic engine mount designs.

1.2

The passive hydraulic mount, illustrated in Figure 1.3, consists of two fluid-filled chambers connected through a decoupler and inertia track. Typically, the fluid within the mount is a mixture

Fo(t)

m

x(t)

FT(t)

z=0.1

desired

operating

range

z=0.3

z=0.6

0

0

0.5

1.5

2.5

3.5

(a)

(b)

Figure 1.2: High frequency excitations: (a) SDOF forced input excitation; (b) transmitted force

frequency domain plot.

of ethylene glycol and water. The upper chamber is bound on top by the main rubber compliance

structure and on the bottom by a steel separator plate which houses the inertia track and decoupler. Under operating conditions the upper compliance holds the static weight of the engine and

as excitations are applied, a fluid pumping action occurs pushing fluid between chambers. The

separator plate, between the two chambers, is fixed to the base of the mount through a rigid

structure that surrounds the lower chamber. The lower chamber is bound by a compliant rubber

bellow, or lower compliance, that expands and contracts as fluid passes though the inertia track

and decoupler.

The decoupler plate has a finite travel distance within its cage. This device limits the

volume of fluid that can pass relatively freely between the upper and lower chambers. Once the

plate bottoms out on the cage the fluid must pass through the inertia track, a long column of fluid

between the two chambers. During small amplitude excitations the fluid passes freely through the

Upper

Compliance

Upper

Chamber

Decoupler

Decoupler

Cage

Lower

Chamber

Inertia

Track

Lower

Compliance

decoupler into the more compliant lower chamber, giving the mount low damping and stiffness

characteristics. Fluid is forced through the inertia track during large amplitude vibrations, as

the decoupler bottoms out in its cage. The resistance and mass of fluid within the inertia track

generate increased stiffness and damping characteristics under these conditions.

1.2.1

Typically, hydraulic mounts are isolated and evaluated in the frequency domain using a servocontrolled hydraulic rate machine. Figure 1.5 illustrates the MTS 1000-Hz rate machine used

to determine the mount characteristics. During the test, the hydraulic actuator

1 is controlled

using a LVDT to excite the mount with a sine wave at a predetermined amplitude and frequency.

In addition, a static preload, or mean force, is applied to represent the static mass of the engine.

The transmitted force at the mount base is measured using a force transducer

2 . These readings

are used as feedback to control the mean force applied to the mount and to measure the amplitude

and frequency of the force applied during frequency excitation.

Dynamic Characterization software [22], the excitations are applied for a duration of 30 seconds

Servo-controlled

Actuator

Hydraulic

Mount

Force

Transducer

LVDT

Controller

Command

A/C Conditioner

D/C Conditioner

Data Acquisition

Workstation

Hydraulic

Actuator

1

Hydraulic

Mount

3

Force

Transducer

2

Figure 1.5: Illustration of MTS 1000-Hz rate machine used to measure dynamic characteristics.

for each frequency and amplitude condition. The program logic requires that each amplitude,

frequency and preload fall within a user described tolerance; 5% in all tests conducted during

this work. At the end of the frequency dwell, once the desired amplitude and frequency are

achieved, the high speed data acquisition samples both excitation X, and transmitted force FT ,

for a duration of time based on the frequency. Sine Regression is applied to each data sequence

to determine the amplitude and phase, as illustrated in Figure 1.6.

Imaginary

Aforce

Fforce

Real

Fexcitation

Aexcitation

These vectors are used to calculate the cross point dynamic stiffness K and phase of the

mount, as

Af orce

Aexcitation

(1.6)

= f orce excitation

(1.7)

K =

The process is repeated for each test condition to develop the frequency domain system

characteristics. Between each frequency excitation the actuator is stopped for a mean dwell,

allowing the force to settle back to the constant preload Fp and resetting all internal dynamics of

the mount to zero.

1.2.2

Mount Characteristics

The decoupler gives the hydraulic mount its desired amplitude dependent characteristics; however, these mounts also exhibit liquid column resonance. Since the columns of fluid residing

between the inertia track and decoupler are not directly connected to the input excitation, they

generate additional degrees of freedom within the system. At some input frequency the liquid column, connected between the upper and lower chamber, will reach its natural frequency,

thus creating the liquid column resonance and altering the hydraulic mount characteristics. Figure 1.7a illustrates how the liquid column resonance of the inertia track influences the overall

dynamic characteristics of the fluid mount. This typically occurs between 5 to 25 Hz and causes

Phase (Deg)

Phase (Deg)

Frequency (Hz)

Frequency (Hz)

(a)

(b)

Figure 1.7: Typical mount dynamic characteristics: (a) Low frequency, large amplitude; (b)

High frequency, small amplitude.

Since there is also a mass of fluid associated with the decoupler, another liquid column

resonance in the high frequency range exists. Figure 1.7b, illustrates a typical high frequency

dynamic stiffness curve which includes the resonance of the decoupler. The high frequency

10

resonance of the decoupler mass is an undesirable effect, since it increases the dynamic stiffness

at high frequencies.

The characteristics of a hydraulic mount are more flexible and desirable than conventional

elastomeric mounts and their use in the automotive industry has become widespread. The trade

off for the enhanced flexibility is increased complexity within the hydraulic mount, and increased

design time. The following survey of papers establishes the current state of research in the area

of hydraulic mount modeling, which will clarify the focus of this thesis work.

1.3

Literature Review

The first paper revealing hydraulic engine mounts is by Bernuchon [1] in 1984.

The au-

thor presents the physical properties of a hydraulic mount and experimentally characterizes the

mounts dynamics. This paper is the first to introduce the excitation conditions and techniques

for dynamic stiffness and phase measurement. Simplified models are documented, but no mathematical development is given. Low frequency characteristics are related to liquid column resonance of the inertia track and high frequency experimental results are plotted up to 200 Hz. The

author then uses on-vehicle experimental results to conclude the superiority of hydraulic mounts

over conventional. In the same year, Corcoran and Ticks [4] describe hydraulic mounts and give

insight into the key characteristics. The influence of decoupler gap and inertia track geometry

is investigated using experimental dynamic stiffness and phase results from 1 to 40 Hz. A vehicle analysis using hydraulic engine mounts shows a 5dB improvement in the noise level at the

drivers ear for a 6 cylinder engine at 3300 RPM. The hydraulic mount is also shown to improve

engine bounce from road excitations. To conclude, the authors make recommendations on how

to apply hydraulic mounts to an existing vehicle.

Flower [5] builds on the understanding of hydraulic mounts by providing physical and

mechanical system models.

11

main mathematical expressions for the cross point dynamic stiffness. The models are the first

to associate an inertia with the decoupler, which is evident during high frequency low amplitude excitations. However, no experimental results are used to validate simulation results. In

the work by Ushijima et al. [32] the dynamic characteristics of a hydraulic mount are studied

in response to superimposed inputs. Experimental evidence is used to show that a compliant

decoupler provides improved performance to superimposed inputs by softening the strong decoupler nonlinearity. The authors developed a hydraulic mount which can achieve low dynamic

stiffness up to 800 Hz, by utilizing liquid column resonance and rubber surging. Also included

in the paper is an investigation into a semi-active mount incorporating electro-rheological fluids.

Seto et al. [29] provide a detailed analysis of the hydraulic mount and predict ideal performance

conditions. Design optimization theory for a dynamic absorber is applied to a mathematical

model of a hydraulic mount with only an inertia track. A simplified mechanical model is developed and the dynamic equations are derived analytically for use in optimization. A conditional

expression for minimizing engine vibration gives the optimum values of fluid mass and damping

of the inertia track. This work factors engine mass into the mount analysis and uses experimental results to validate the design method proposed. Further work focused on the inertia track is

conducted by Gau and Cotton [6]. The authors propose two categories of hydraulic mounts, single and double-load-bearing fluid chambers, and review mathematical models for both. A linear

lumped parameter model is developed for a standard single-load-bearing chamber mount with

only an inertia track, and results are compared to experimental data. The paper compares model

to experimental results in the low frequency range and investigates the influence of device parameters. The detailed study considers chamber compliance, inertia track area and length, and

fluid viscosity. A process for hydraulic mount design is presented, then a prototype fluid mount

is tuned to reduce vibration in an engine system.

12

The first of many papers by Singh et al. [30] represents a culmination of all work cited

to this point.

Linear, time-invariant, lumped parameter models for both free and fixed type

decouplers are mathematically developed. By splitting the hydraulic mount model into two, large

amplitude low frequency and small amplitude high frequency, the decoupler nonlinearity can be

avoided. The authors show that most prior models are special cases of their own. Mechanical

system models are developed, however they do not capture the transmitted force accurately.

Models are also presented in state-space form and validated over 1 to 50 Hz.

The authors

are the first to identify a problem with the hydraulic mount during high frequency excitations.

Experimental results show increased dynamic stiffness beyond 150 Hz and poor correlation with

the linear model, however the issue is not investigated further. The paper concludes with low

frequency parameter studies on the simplified model.

Branch and Haddow [20] provide a more fundamental explanation of the hydraulic mount,

using a previously developed linear model. This model is used to interpret the dynamic stiffness

curve by observing zeros and poles in the Laplace domain equation.

validation is also conducted with the experimental data. A thorough review of engine vibration

sources and corresponding frequency ranges is also developed and used to explain desired engine

mount characteristics. Lee et al. [15] present modeling and performance analysis of a hydraulic

engine mount using the bond graph method. Model validation is shown over 1 to 50 Hz and 1 to

100 Hz ranges using experimental data.

More recent work begins to focus on the nonlinear dynamics of hydraulic mounts. Kim

and Singh begin an extensive nonlinear analysis of the hydraulic mount. In [12], the nonlinear

properties are identified for a fluid mount with an inertia track.

By experimentally measur-

ing the chamber volumetric compliance, an equation is fitted to the static curve, characterizing

a nonlinear chamber compliance. A nonlinear form of the fluid resistance within the inertia

track is also determined experimentally. A nonlinear lumped parameter model is formulated

13

and compared with experimental data over time and frequency domains from 1 to 50 Hz. Discrepancies between theory and experiment are attributed to gas-liquid phase transformation and

cavitation phenomenon. The content within their next two papers [11, 13] examines the dynamic

characteristics of a hydraulic mount in an isolated case and within a simplified vehicle model.

The nonlinear lumped parameter model is updated to include decoupler switching characteristics. Using the pressure differential and volume of fluid passing through the decoupler, flow is

switched by effectively increasing the pressure differential within the decoupler. Extensive experimental verification of the nonlinear model includes harmonic response data collected in the

time and frequency domain. To examine the vibration isolation and shock absorption properties,

simplified vehicle model experiments are conducted and include both harmonic and impulse excitations.

50 Hz. However, the high frequency decoupler resonance shown experimentally is again not

pursued. The authors work also includes a new broadband adaptive hydraulic mount. Most

work documented in the previous three papers are a part of Kim [10].

Colgate et al. are the first to address the high frequency decoupler resonance [3]. Both

low and high frequency conditions are investigated as well as composite input response. The

authors conclude that high frequency isolation is significantly degraded when combined with a

large amplitude low frequency disturbance. Using two linear models for coupled and uncoupled

decoupler conditions, a piecewise linear simulation technique is used to show an amplitude dependent frequency response. State space and frequency domain equations for both linear models

are developed using the bond graph method. Model parameters are identified using the dynamic

stiffness curves of specially designed mounts, and the decoupler inertia is characterized using

high frequency experimental results. The piecewise decoupler model switches between integrating large and small amplitude equations depending on the displacement of the decoupler and

the decoupler velocity. Time domain simulations are conducted, however no results are shown.

14

The simulations represent experimental data over the complete low and high frequency ranges

of interest, however the authors state that moderate improvements can be attained by capturing

the effects of the lower chamber compliance, upper chamber bulge damping and leakage past

the decoupler. Margolis and Wilson [18] present a nonlinear model of the hydraulic mount that

uses first principle representation for inertia track resistance and decoupler. Momentum of the

decoupler is stopped by applying an additional resistive force, dependent on the position of the

decoupler and the pressure differential.

validated using experimental data, however high frequency results do not correlate well.

The final set of papers places more emphasis on vehicle vibration with hydraulic mounts.

Ushijima and Dan [31] use the nonlinear building block approach to predict vibration of a vehicle

with hydraulic engine mounts. The authors combined the numerical simulation of a hydraulic

mount with modal analysis of a vehicle. In [8], Ishihama et al. present a new design method for

tuning low frequency vehicle vibration problems. The work uses engine and body modal analysis to determine mount contribution to vibration. Phase angle control on a hydraulic mount is

then used to adjust vibration levels within the vehicle. Inertia track damping is used to alter the

dynamic stiffness and phase angle at desired frequencies. The paper by Muller [23, 24] gives

design strategies for mount systems based on calculated and experimental methods. A rigid

body model of the engine and mount system is used to aid mount design for quasi-static and dynamic loads. FEA (Finite Element Analysis) tools are illustrated to be effective in predicting

axial stiffness, volumetric compliance and effective pumping area of a hydraulic mount. The authors present a standard and switchable hydraulic mount as well as a hydraulic bushing, but only

physical details and limited experimental results are shown. The strategies of transfer path analysis are explained and illustrated. Finally, the work investigates an active vibration absorber. In

the three papers by Royston and Singh [26, 27, 28], an extensive analytical analysis is developed

for nonlinear mounting systems based on the computational Galerkin method. Most of the work

15

presented is focused on the response of a simplified engine and nonlinear mount system and all

nonlinear hydraulic mount modeling is based upon previous work by Kim and Singh. Little and

Kashani [16] also use existing models of hydraulic mounts to focus on decoupler switching characteristics. The authors apply smart materials such as magnetostrictive and electro-rheological

fluid to the flow communicating through the decoupler, to achieve semi-active and active hydraulic mounts.

1.4

To reduce the time required to design a hydraulic mount, it is desirable to construct fluid mount

models that can be simulated to predict the system response before it is physically assembled.

From the preceding literature review, the shortcomings in hydraulic mount modeling are mainly

in the high frequency range. Colgate et al. [3] developed a model that is effective at high frequencies, however it is based on a piecewise linear approach.

physically intuitive model that is effective over the complete excitation range of interest. Furthermore, it is important for designers to understand how features within the hydraulic mount can

be changed to achieve the desired mount characteristics. Parameter investigations in the high

frequency range are also not presented in any literature reviewed.

This thesis consists of a complete guide to nonlinear hydraulic mount modeling and simulation, in an effort to reduce hydraulic mount development time.

Chapter 2 consists of an

extensive system modeling portion, that first redevelops the best models reviewed in the literature. Enhancements to the nonlinear decoupler model include a continuous function that follows

a simplistic, yet effective, approach to capture the switching effect. Leakage through the decoupler is considered and a clear nonlinear model of the transmitted force is shown. Upper chamber

bulge damping has also been accounted for in the system model. Resistance to oscillating fluid

flow is accounted for by using a simple equation that can be related to first principle fluid dy-

16

namic equations. Finally, the modeling covers a unique bell component within the hydraulic

mount that alleviates the high frequency liquid column resonance of the decoupler.

Chapter 3 is dedicated to providing a physical explanation of low and high frequency characteristics. Time domain plots of the decoupler operation give an indication of how this component influences the system. Liquid column resonance is explained and related to system output.

In addition, the high frequency function of the additional bell component is presented in detail.

This portion also gives an indication of how each lumped parameter effects the overall system

characteristics.

Perhaps the most unique portion of this work is the experimental process, outlined in Chapter 4. This thesis covers a method whereby components of the hydraulic mount are isolated to

enable parameter identification. Chapter 4 introduces the method and apparatus, and documents

the calibration procedure.

Details of component parameter identification are given in Chapter 5. This includes the

test setup, perturbation conditions and system identification approach for each component.

The system simulation accounts for all experimentally determined nonlinearities within the

system, and is documented in Chapter 6. Experimentally measured parameters are incorporated

into a nonlinear simulation algorithm and compared with complete hydraulic mount system test

data.

Preliminary findings for parameter to geometry relationship are outlined in Chapter 7. This

brief section includes the initial findings for the relation between parameters identified and used

in the system simulation, and the geometry of the components.

Chapter 8 concludes the thesis. The key findings are highlighted and recommendations

for future work in this area are made.

Chapter

All mathematical models used in this thesis to represent hydraulic mount dynamic behavior are

developed in this chapter. First, a linear model of a typical hydraulic mount is derived and compared with previous work in this area.

while mechanical models are used to illustrate mount behavior. The chapter focuses on developing a nonlinear model that is effective across the frequency and amplitude ranges of interest;

the decoupler model represents an advancement in hydraulic mount modeling. Finally, a bell

system within the mount is presented and modeled. This additional component within a hydraulic mount is used to alleviate the high frequency resonance of the decoupler. To the best of

the authors knowledge, this system has not yet been modeled in any literature.

2.1

Linear models for a typical hydraulic mount are developed using the bond graph method. Lumped

parameters and variables are first assigned to physical characteristics, then the bond graph method

is applied to determine the linear system equations. The system equations are then divided into

two linear models, thereby providing a better representation of the actual system in both high

and low frequencies.

17

18

Upper

Compliance

C1 Ap kr br

Decoupler

Id R d

Upper

Chamber

Inertia

Track

Ii Ri

Lower

Chamber

Lower

Compliance

C2

2.1.1

Consider the hydraulic mount cross section illustrated in Figure 2.1. The upper chamber compliance serves three main functions within the system.

stiffness and damping properties of the mount kr and br , respectively. The upper compliance

also functions as a piston, with an effective pumping area Ap . Finally, the rubber structure adds

volumetric compliance to the system, represented by C1 . The inertia track, a long column of

fluid between chambers, is assigned lumped parameters Ii and Ri representing the inertia and resistance, respectively. The other flow passage between the upper and lower chambers is through

the decoupler. To begin constructing a system model, the decoupler is assigned linear lumped

parameters Id and Rd , which also represent an effective inertia and resistance. Assigning the

decoupler an effective inertia is key to establishing high frequency characteristics and has been

included in [3, 5, 15, 32].

ance and is modeled using a lumped parameter C2 . Using the assigned parameters, the system

schematic is converted to a lumped parameter model, illustrated in Figure 2.2.

19

X(t)

kr

br

P1(t)

Qi(t)

Qd(t)

C1

XT(t)

Ri Ii

Rd Id

FT(t)

P2(t)

C2

Ap

Variables within the system include the input excitation X(t) and motion of the mount base

XT (t). Also, the upper and lower chamber pressures are captured by P1 (t) and P2 (t), respectively. Flow through the inertia track Qi (t) and decoupler Qd (t) are highlighted in Figure 2.2,

as well as the transmitted force to the mount base FT (t).

2.1.2

The bond graph method is applied to develop equations of motion from the lumped parameter

model. This method maps the flow of energy within the hydraulic mount (see Appendix A or

van den Bosch et al. [2] and Karnopp et al. [25]) and is particularly useful for multi-disciplinary

systems. Since the hydraulic mount system is a combination of fluid and mechanical components, the bond graph method is applied as represented in Figure 2.3. This method has also been

applied in [3, 5, 15, 18] to determine the linear state equations.

The

Input excitation at the top of the mount is represented by a source of flow variable X.

effective piston area is now represented by the transformer variable and each compliance parameter is connected to pressure nodes of the upper and lower chamber. Variables and parameters in

20

:1/kr

: Ap

: X(t)

: br

: C1

: Ri

: Rd

: Qi(t)

: Qd(t)

: Id

: Ii

: C2

Figure 2.3: Bond graph model for a typical hydraulic engine mount.

the bond graph model represent those used in the lumped parameter system. However, since all

experimental studies are conducted with a fixed base test fixture, the base motion variable XT is

set to zero.

2.1.3

System Equations

Using the bond graph model, the equations for the internal dynamics of the system are established. The continuity equations become

C1 P1 = Ap X Qi Qd

(2.1)

C2 P2 = Qi + Qd

(2.2)

P1 P2 = Ii Q i + Ri Qi

(2.3)

P1 P2 = Id Q d + Rd Qd

(2.4)

The system of four state variables and one input, represents the internal dynamics of a fluid

mount. The transmitted force developed using Figure 2.4 is dependent on the decoupler position,

however, a nonlinear function for this dependence will be derived later. Two situations for the

Ap

21

Ap

X(t)

X(t)

kr

br

P1(t)

Qd(t)

P2(t)

kr

br

Rd

FT(t)

P1(t)

Qd(t)

Rd

P2(t)

FT(t)

Ad

(a)

(b)

Figure 2.4: Transmitted force to the mount base: (a) decoupler closed; (b) free floating decoupler.

decoupler arise and can be considered separately, similar to the work in [3, 13]. For Figure 2.4a

the decoupler is contacting the cage and the transmitted force equation is

FT = kr X + br X + Ap (P1 P2 ) + Ap P2

(2.5)

FT = kr X + br X + (Ap Ad )(P1 P2 ) + Ap P2 + Ad Rd Qd

(2.6)

In both cases the force transmitted to the mount base represents the sum of forces applied

through the mechanical stiffness and damping parameters, along with the influence of internal

chamber pressures.

In (2.5) the upper chamber pressure acts on the mount base and piston

area, while the lower chamber pressure is internal and does not influence FT . In (2.6) the upper

chamber pressure acts over the same area, however it is now reduced by the decoupler hole. Also,

the flow resistance through the decoupler influences the total transmitted force, captured initially

with Ad Rd Qd (similar to Colgate et al. [3]). Equations (2.1) to (2.6) can now be separated into

models that are dependent on the excitation amplitude; capturing the nonlinear behavior of the

hydraulic mount with two linear models.

2.1.4

22

A linear approach to modeling hydraulic mounts can be used if small and large excitation amplitudes are considered as independent models, this follows a similar technique used by Singh et

al. [30] and Colgate et al. [3].

The Small Amplitude High Frequency Model assumes that during small amplitude excitations the decoupler does not contact the cage and all flow passes freely through the decoupler.

Also, the high frequency excitations are generally above the liquid column resonance of the inertia track, hence negligible flow passes through the inertia track. A linear model that captures the

system response for high frequency small amplitude excitations is developed by removing equation (2.3), and state variable Qi , from the state equations previously generated. The continuity

equations for the small amplitude high frequency become

C1 P1 = Ap X Qd

(2.7)

C2 P2 = Qd

(2.8)

P1 P2 = Id Q d + Rd Qd

(2.9)

For this model, the decoupler is assumed to be free at all times, therefore Figure 2.4b applies and

equation (2.6) is used to represents the transmitted force.

FT = kr X + br X + (Ap Ad )(P1 P2 ) + Ap P2 + Ad Rd Qd .

(2.10)

The Large Amplitude Low Frequency Model assumes all flow passes through the inertia

track, since during large amplitude excitations the decoupler will spend the majority of time

against the cage with no flow passing through the decoupler. The equations for the linear model

are developed by removing the fluid momentum equation (2.4) from the system of equations

23

C1 P1 = Ap X Qi

(2.11)

C2 P2 = Qi

(2.12)

P1 P2 = Ii Q i + Ri Qi

(2.13)

From Figure 2.4a, the decoupler is assumed fixed and thus (2.5) is the transmitted force equation.

FT = kr X + br X + Ap (P1 P2 ) + Ap P2

(2.14)

Using these linear system equations, equivalent mechanical system models can be developed to

illustrate the physical operation of the hydraulic mount.

2.2

The linear system equations are now used to develop the mechanical model. The main area of

interest is the fluid column dynamics, since the stiffness and damping properties kr and br , are

already represented as a mechanical system. First consider the fluid system equations (2.7) to

(2.9) for the high frequency small amplitude model,

C1 P1 = Ap X Qd

C2 P2 = Qd

P1 P2 = Id Q d + Rd Qd

Substituting (2.7) and (2.8) into the derivative of (2.9) gives,

d + Rd Q d +

Id Q

1

1

+

C1 C2

Qd =

Ap

X

C1

(2.15)

24

The system parameters and variables are converted to their equivalent linear mechanical form by

considering the ideal fluid system elements (see Rowell [25])

Qd = Ad X d

md

A2d

bd

=

A2d

(2.16)

Id =

Rd

In (2.16) md , bd , Ad and X d are the effective mass of the decoupler, the damping coefficient for

the linear decoupler motion, the area of the decoupler, and the linear velocity of the decoupler,

respectively. The compliance parameters are transformed from volumetric stiffness to a linear

stiffness using,

A2p

A2p

C1 =

, C2 =

k1

k2

(2.17)

where k1 and k2 are linear stiffness parameters of the upper and lower compliance, respectively.

Substituting (2.16) and (2.17) into (2.15) and integrating with respect to time, gives

d + bd X d + (k1 + k2 )

md X

A2d

Ad

X

=

k

X

d

1

A2p

Ap

(2.18)

with all initial conditions set to zero. Recall that (2.18) only represents the internal dynamics

of the fluid mount. To construct the complete mechanical system the additional stiffness and

damping variables of the upper chamber must be included. Figure 2.5 illustrates the mechanical

system model for the small amplitude high frequency model, following the same techniques as

Flower [5].

Before proceeding, the transmitted force to the base is derived for the mechanical model to

ensure this system is mathematically equivalent to the original system equations. Observing the

transmitted force at each pin in Figure 2.6, the force equations are determined.

25

Figure 2.5: Mechanical representation of the high frequency, low amplitude linear model.

26

FT _1 = kr X + br X

FT _2 = bd X d

(2.19)

Ad

Xd

Ap

Ad

Ad

d

= k1 X

Xd k2 Xd bd X d md X

Ap

Ap

FT _3 = k2

FT _4

These equations are then rewritten in terms of the fluid system parameters using the transformations in (2.16) and (2.17) and also using the relationships in (2.7) to (2.9):

FT _1 = kr X + br X

FT _2 = Rd Ad Qd

(2.20)

FT _3 = Ap P2

FT _4 = (Ap Ad ) (P1 P2 )

The summation of the transmitted force equations in (2.20) is

FT = kr X + br X + (Ap Ad ) (P1 P2 ) + Ap P2 + Ad Rd Qd

(2.21)

Comparing equation (2.6) with (2.21) validates that Figure 2.5 corresponds to the small amplitude high frequency model. Similar work was presented by Flower [5], however no mathematical

validation was given.

The next logical extension is to incorporate the inertia track into the mechanical model.

However, with both the decoupler and inertia track the model becomes nonlinear. To illustrate

the concept, Figure 2.7 includes an inertia track mass with a switch that engages the inertia track

when the decoupler motion reaches a certain displacement. This approach has also been presented in [5]. Also, the damping element on the inertia track is not connected to the ground

where transmitted force is calculated since the inertia track travels perpendicular to the direction

of transmitted force. The nonlinear mechanical model is developed solely for illustrative pur-

27

Figure 2.7: Mechanical model that includes the decoupler and inertia track.

poses and is not studied any further in this thesis. Instead, effort will be placed on enhancing the

fluid system model to capture nonlinear effects.

2.3

This portion of the hydraulic mount modeling is devoted to the development of a nonlinear

lumped parameter model that will represent the mount response over the complete excitation

range. The nonlinear model evolves by enhancing the parameters of the linear model developed

in Section 2.1. These enhancements capture nonlinearities in the upper chamber stiffness, effective pumping area, chamber volumetric compliance, inertia track resistance, decoupler switching,

transmitted force, and decoupler leak flow.

The model enhancements have been developed in conjunction with component evaluations,

conducted experimentally and documented in Chapter 5. To the best of the authors knowledge,

this section represents an advancement to the nonlinear modeling work of Kim and Singh [11, 13]

and Colgate et al. [3].

2.3.1

28

Upper Chamber

Stiffness and Damping In the preceding sections the variables kr and br have been introduced to

represent the stiffness and damping properties within the upper chamber. In all papers reviewed,

these variables are treated using the Viogt model for rubber, which assumes both properties are

invariant with frequency and amplitude. In an effort to develop a complete non-linear model,

more attention has been given to associating the variability of these parameters with the frequency and amplitude of excitation. Since methods are unavailable to reliably predict dynamic

characteristics of rubber from physical geometry, the characteristics will be gathered experimentally and identified in Chapter 5. The behavior of system parameters kr and br in the frequency

domain are shown to be dependent on the steady-state amplitude and frequency of the excitation,

as well as the preload force magnitude. Stiffness and damping are written in the functional form

Fp

kr wdr , X,

(2.22)

Fp )

br (wdr , X,

where wdr represents a frequency of oscillation, X

Effective Pumping Area The effective pumping area is dependent on the geometry of

the rubber, but the parameter cannot be easily determined from the geometry. Under the static

load of the engine mass, the rubber is deformed prior to the excitations. This preload force

Fp yields a mean displacement of the rubber that influences the effective pumping area. The

effective pumping area Ap is a function of the preload displacement and has been experimentally

measured in Chapter 5. Since the preload displacement is directly related to the static force,

through the static load displacement curve, the effective pumping area is written as a function of

the preload force

Ap (Fp )

(2.23)

29

Compliance The variable C1 represents the volumetric compliance of the upper chamber,

however it does not capture the hysteresis that is evident in the volumetric expansion. Hysteresis

is a result of molecular chains unraveling as the material expands and contracts, and is shown

experimentally to be evident in the expansion of the upper chamber in Chapter 5. The net effect

on the system response is an apparent increase in damping.

that models are moderately improved by including bulge damping effects that are the result of

hysteresis. In the fluid system representation the volumetric damping is captured by including a

resistance parameter on the flow into the compliant region.

Figure 2.8 illustrates the changes from the standard modeling technique.

Figure 2.8a

shows a fluid system representation of the upper chamber compliance used in previous models,

while Figure 2.8b illustrates how the bulge damping has been modeled with the addition of a

flow resistance. A bond graph of this model is illustrated in Figure 2.9 and is used to develop

the equation

C1 P1 = QT + C1 R1 Q T

P1(t)

(2.24)

P1(t)

C1

Qt(t)

C1

Qt(t)

R1

(a)

(b)

Figure 2.8: Capturing bulge damping in the upper chamber: (a) previous model; (b) addition

of a resistance term in the compliance.

Here, QT has been introduced to represent flow entering into the compliant region and the

parameter R1 represents the flow resistance. The variable P1 is the same upper chamber pressure

used in previous models.

30

:R1

QT(t)

:P1(t)

: C1

Figure 2.9: Bond graph model of the compliance system with resistive parameter.

The same issues arise with the parameters C1 and R1 , as with the mechanical stiffness and

damping properties. These parameters are dependent on the steady-state volume amplitude VT

and the frequency wdr of excitations. Due to the geometrical changes under static preload, the

nonlinear parameters are also considered as functions of the preload force Fp

C1 (wdr , VT , Fp )

(2.25)

R1 (wdr , VT , Fp )

These parameters are determined experimentally in Chapter 5.

2.3.2

Lower Chamber

The influence of the lower chamber compliance on the mount characteristics is, to a large extent,

dependent on the system design. Hydraulic mounts that use the lower chamber as a load bearing

chamber will generally have stiff lower compliance values that significantly influence the system dynamics. If this is the case, the volumetric parameters should be assigned to capture the

compliance and bulge damping characteristics, similar to (2.25). However, if the main function

of the lower chamber is to accommodate fluid transfer, the volumetric compliance will have little influence on the system dynamics. Under these conditions a single volumetric compliance

parameter C2 is used to characterize the lower compliance. This design approach is the most

common and is the one considered in this work.

31

Under certain circumstances the volume of fluid pushed into the lower chamber during

preload will increase the chamber compliance parameter as the rubber stretches. Depending

on the design, it may be appropriate to model the lower chamber compliance as a function of

the static preload displacement, which is related to the preload force through the static load

displacement curve,

C2 (Fp )

2.3.3

(2.26)

Inertia Track

The linear inertia track model assumes a constant inertia Ii and resistance Ri . To enhance this

model, some fluid dynamic properties are considered. Kim [10] has concluded that the fluid

density can be considered constant and that the thermodynamic effects on the flow are negligible

for the oscillatory flow in an inertia track. However, Kim and Singh [13] have used an increased

fluid inertia for laminar flow within the inertia track. All other papers regarding the inertia track

dynamics assume a constant inertia parameter. Using previous research and experimental results

detailed later, the inertia parameter is concluded to be a constant parameter in the model.

Unlike the inertia parameter, the resistance does not exhibit constant behavior. To determine the flow resistance trends within the inertia track, this work first considers the fluid

equations for losses through a pipe (see White [33, pg.363]) by

P = f

8L 2

8 2

Q

+

k

Q

2 d5

d4

(2.27)

where L and d are the length, and diameter of the pipe and is the fluid density. Also,

f d , RED and k (d, D) represent friction factor coefficients for a pipe and entrance losses, respectively. These coefficients are dependent on the Reynolds number (RED ), the mean surface

32

Equation (2.27) can be used to model continuous fluid flow into a pipe with some reasonable degree of accuracy, however the inertia track flow is oscillatory. Since the flow is sinusoidal,

the Reynolds number will change at each point within a period of flow oscillation. As the flow

is changing direction the Reynolds number will approach zero, then it will increase to a maximum, passing through the laminar and into the turbulent region. The amount of time spent in

the laminar and turbulent regions will be dependent on the amplitude and frequency of the flow.

Since both f and k will fluctuate as the flow oscillates, a mean value for each is used. The flow

variable Q, now written as the inertia track flow variable Qi , is factored out of (2.27) and an effective resistance value Ri2 is defined. This is combined with the existing resistance parameter

Ri in (2.3), now called Ri1 , to form the nonlinear resistance equation

(2.28)

The inertia term is now combined to provide a nonlinear fluid momentum equation from (2.3).

P1 P2 = Ii Q i + (Ri1 + Ri2 |Qi |) Qi

(2.29)

In (2.28) and (2.29) it is necessary to place the flow variable Qi inside absolute value signs

to achieve the appropriate influence on the pressure differential.

2.3.4

Decoupler

Flow Control The nonlinear switching behavior of the decoupler accounts for most of the amplitude dependent characteristics.

the decoupler plate does not contact the cage, the decoupler is considered open and the system

behaves as a simple orifice. In equation (2.4) the decoupler is assigned a lumped inertia and resistance value similar to the inertia track. The inertia of the decoupler exists, and is considered

to include the decoupler plate and a column of fluid above and below. As with the inertia track,

33

the decoupler inertia is assumed constant with amplitude and frequency of flow excitation and is

experimentally validated later.

The free-floating decoupler resistance is analyzed in a similar fashion to the inertia track.

Starting with the equations for fluid loss through an orifice (see White [33, pg.363])

P =

2

Q

A2d

(2.30)

where is the fluid density and Ad is the decoupler orifice area. Also, (, RED ) is an orifice

friction factor which is a function of the Reynolds number RED and the velocity-of-approach

factor . Again, the flow is oscillatory and thus the Reynolds number does not remain constant.

Using the same oscillatory flow rationale used for the inertia track, the nonlinear momentum

equation for the free floating decoupler becomes

P1 P2 = Id Q d + (Rd1 + Rd2 |Qd |) Qd

(2.31)

where the parameters Rd1 and Rd2 represent the new resistance parameters.

Equation (2.31) applies to the case where the decoupler plate is free from contacting its

cage. When the decoupler plate does contact the cage it blocks all flow across the decoupler

and introduces the amplitude dependent nonlinear characteristics. The additional flow-stopping

effect can be mathematically represented by making the decoupler orifice area approach zero,

Ad 0, in equation (2.30), which drives Qd 0.

relatively large resistance term to equation (2.31). The additional resistance term is captured

with Radd and is applied to equation (2.31) to give

P1 P2 = Id Q d + (Rd1 + Rd2 |Qd | + Radd ) Qd

(2.32)

Since the flow stopping resistance Radd is a nonlinear parameter, the switching characteristics must be established. In [11, 13, 18] the decoupler switching characteristics are dependent

on the volume of fluid passed through the decoupler and the pressure differential between upper

34

and lower chambers. However, the work for this thesis has established that the switching characteristics are dependent on the volume of fluid and the direction of flow across the decoupler.

During low frequency excitations this differentiation will have no influence on the nonlinear response, since the low inertia decoupler column moves in phase with the pressure differential.

However, at high frequencies, typically above 100 Hz, the inertia effect will cause the decoupler

column to oscillate out of phase with the pressure differential.

Using the volume of fluid and the flow direction across the decoupler as switching variables, a mathematical representation is now established. A nonlinear function for Radd is developed using the decoupler position in an exponential function with the arctangent function to

model the flow dependent behavior.

The flow of fluid through the decoupler Qd is integrated to obtain the decoupler volume

Z T

Vd =

Qd dt

0

with the decoupler initially positioned on centre at time zero. The decoupler position Xd is

related to the volume via the measured decoupler area Ad .

Xd =

Vd

Ad

(2.33)

As the decoupler contacts the cage limits, the resistance Radd takes effect and reduces

the decoupler flow to effectively zero. When the flow reverses direction, a result of the input

pressure differential and momentum within the system, the additional resistance is removed as

the decoupler travels relatively freely to the opposing cage limits. Figure 2.10 illustrates the

decoupler sign convention and the desired function of Radd .

An exponential function is developed to model the desired additional resistance term or

flow-stopping action shown in Figure 2.10.

Radd = 1 eRd3 Xd arctan(Qd 2 )

(2.34)

35

Radd

-Qd(t)

+Qd(t)

+Qd(t)

-Qd(t)

Qd(t)

0

Xd

Xd

(a)

(b)

Figure 2.10: The nonlinear decoupler resistance: (a) sign convention; (b) switching function.

where 1 is a constant used to place the overall additional resistance magnitude in the appropriate

range.

position at which the exponential increases and the additional resistance reduces decoupler flow.

Figure 2.11 indicates how an increase in Rd3 is comparable to closing the decoupler gap.

Flow dependent characteristics of equation (2.34) are developed by switching the exponential sign using the principle branch of the arctangent function (arctan). This function is applied

to the decoupler flow to achieve a continuous output of 1, dependent on the direction. The

constant 2 is included to adjust the shape of the function output. To demonstrate how 2 was

selected, consider the arctangent function in isolation

2

arctan(Qd 2 )

(2.35)

Figure 2.12 illustrates the impact of constant 2 on the shape of this function. To achieve a crisp

switching response the constant 2 is set to 1 105 s/mm3 for all cases. The

in (2.35) is

The final form of the momentum equation (2.31) becomes

(2.36)

36

-7

x 10

4.5

4

3.5

3

2.5

2

1.5

1

0.5

-0.45

-0.30

-0.15

0.15

0.30

0.45

Rd3 = 35

Rd3 = 32

Rd3 = 28

1

0.8

0.6

0.4

0.2

0

-0.2

-0.4

-0.6

-0.8

-1

-100

-80

-60

-40

-20

20

40

60

80

100

x2 = 1x105

x2 = 1

x2 = 1x10-1

37

This equation represents a nonlinear function that captures the decoupler momentum and introduces one new parameter to the model Rd3 . Equation (2.36) is also used to identify Rd3 from

experimental data, described in Chapter 5.

The transmitted force equations are also influenced by the decoupler system and will be

considered next.

Transmitted Force Improvements in the transmitted force equations (2.5) and (2.6) are

established by including the effects of the cage geometry and switching properties. The transmitted force equations for a typical decoupler and cage model are developed first. Then the

transmitted force equations for the open and closed decoupler situations are combined into one

function through the development of a nonlinear switching function.

The influence of the decoupler cage on the transmitted force equations is modeled using

the uncoupled isolated decoupler system shown in Figure 2.13. The cage geometry includes

the area of the holes Ad_holes and the area of the decoupler plate Ad , while the total area of the

holding plate is denoted by Ap . The pressure differential across the system is represented by the

nonlinear equation (2.36). For the uncoupled case the additional resistance function becomes

negligible and the momentum equation becomes

(2.37)

As Figure 2.13b illustrates, the pressures P1 and P2 are introduced as intermediate pressures within the decoupler and are used to establish the regions of pressure change

P1 = P1 P1

0

Pplate = P1 P2

0

P2 = P2 P2

(2.38)

Ad_holes

P1(t)

38

Qd(t)

P 1(t)

P 2(t)

Ad

P2(t)

Qd(t)

FD

Ap

(a)

(b)

Figure 2.13: Isolated decoupler model that includes cage geometry: (a) geometry parameters;

and (b) pressure gradient breakdown.

where P1 and P2 represent pressure drops across the cage and Pplate denotes the pressure

differential across the decoupler plate. The total pressure across the system is

P1 P2 = P1 + Pplate + P2

(2.39)

Following the lumped parameter approach, the pressure changes P1 and P2 are considered to be only a result of flow through the cage holes and some lumped resistance Rcage . Also,

Pplate is thought to represents a pressure change due to the inertia within the decoupler Id . This

approach assigns the momentum equation to distinct portions of the pressure differential, as

P1 = Rcage Qd

Pplate = Id Q d

P2 = Rcage Qd

(2.40)

(2.41)

(2.42)

Since the resistance across the top and bottom cage are the only contributors to the total

decoupler resistance, P1 and P2 are summed and related to the resistance in equation (2.37).

P1 + P2 = (Rd1 + Rd2 |Qd |) Qd

(2.43)

Substituting (2.41) and (2.43) into equation (2.39) yields the same pressure relationship

established in (2.37), therefore this procedure has not changed the system dynamics.

39

By splitting the decoupler pressure differential, the transmitted force is established using a

force balance. From Figure 2.13b

FD = (Ap Ad )(P1 P2 ) + (Ad Ad_holes )P1 + (Ad Ad_holes )P2

(2.44)

FD = (Ap Ad )(P1 P2 ) + (Ad Ad_holes ) (Rd1 + Rd2 |Qd |) Qd

(2.45)

If the upper chamber stiffness and damping are included, along with the lower chamber pressure,

the total transmitted force becomes

FT = kr X + br X + (Ap Ad )(P1 P2 ) + Ap P2

(2.46)

This equation is developed using the model that assumes the decoupler is open, or free

from contacting the cage. When the decoupler contacts the cage, the transmitted force becomes

equivalent to

FT = kr X + br X + Ap (P1 P2 ) + Ap P2

(2.47)

Two events occur to transform the transmitted force equations from (2.46) to (2.47). First,

the decoupler closes and the pressure differential acts over the full area Ap , essentially Ad 0.

Second, flow across the decoupler becomes negligible, Qd 0.

decoupler is already controlled via the nonlinear decoupler resistance previously developed, only

the decoupler area function still needs to be developed.

The decoupler area is dependent on the decoupler position and pressure differential, following [11, 13, 18]. When the decoupler plate is contacting the lower cage and the pressure

differential P = P1 P2 is positive, Ad will appear to be zero. However, if the pressure differential reverses direction the decoupler orifice will appear open, even if the inertia still holds the

decoupler in place. These switching conditions are different from those used in the flow control

40

Ad_fnc

+DP(t)

-DP(t)

Ad

P1(t)

-Vd_max

0

P2(t)

Vd_max

-DP(t)

-Vd_max

(a)

+DP(t)

0

Vd_max

Vd

(b)

Figure 2.14: The decoupler area: (a) sign convention; (b) switching function.

since the transmitted force is directly related to the pressure differential across, not flow through,

the decoupler.

In the following nonlinear development the parameter Vd_ max is introduced to signify the

maximum volume of fluid accommodated by the decoupler. Using the sign convention illustrated

in Figure 2.14a, the nonlinear Ad_f nc function is illustrated in Figure 2.14b. The conditions for

the decoupler area switching are

Ad_f nc =

0,

Vd Vd_ max ,

Ad ,

Vd Vd_ max ,

0,

Vd Vd_ max ,

Ad ,

Vd Vd_ max ,

P

P

P

P

>0

0

<0

0

(2.48)

To develop a continuous function for Ad_f nc the arctangent function is used on both the

pressure differential and decoupler volume displacement. The function development first includes an arctangent function of the pressure differential multiplied by the volume displacement.

A=

2

Vd arctan ( 3 P )

(2.49)

The constant 3 controls the shape of the arctangent function similar to equation (2.35) and is set

to 1 105 mm2 /N. For positive pressure differentials, the value of A will become Vd . So, when

the pressure differential is negative and the volume displacement is up near the top cage, Vd is

also negative and the value of A will again be positive. By subtracting Vd_ max from A and taking

41

the arctangent of the result, the function will only give a result of +1 when A exceeds Vd_ max .

The function becomes

B=

2

arctan ((A Vd_ max ) 4 )

(2.50)

where a constant 4 is again used to control the shape of the arctangent function, with a value

of 1 103 mm3 . From the conditions on Ad_f nc outlined in (2.48), the value of B will be

+1 for each case where Ad_f nc = 0 is desired and B will take on 1 for each case where

Ad_f nc = Ad is desired. The final step of this function development is to establish an equation

for this relationship.

1

Ad_f nc = Ad (1 B)

2

Writing the nonlinear equation in terms of the pressure differential and volume, gives

2

Ad_f nc = Ad

arctan

Vd arctan ( 3 P ) Vd_ max 4

(2.51)

(2.52)

Finally, the complete nonlinear transmitted force equation from equation (2.46) is

FT = kr X + br X + (Ap Ad_f nc )(P1 P2 ) + Ap P2 +

(2.53)

At this point the nonlinear analysis has captured the dominant switching characteristics

of the decoupler, but an adjustment must be made in the cases where the decoupler does not

completely block the flow.

Leak Flow The nonlinear, decoupler flow control equation completely stops decoupler

flow.

However, some decoupler designs permit high resistance flow after the decoupler has

contacted the cage. The leak flow passes between the decoupler and cage when the decoupler

is in the closed position, thus it only exists when Qd is negligible. For convenience this flow is

referred to as decoupler leak flow, represented by the variable QL . The current decoupler model

is enhanced to capture this leak flow with the addition of an orifice, as illustrated in Figure 2.15a.

42

RL2_fnc

+Qd(t)

-Vd_max

-Qd(t)

+Qd(t)

RL2

QL(t)

Qd(t)

Vd_max

-Qd(t)

RL2 x x6

-Vd_max

(a)

Vd_max

Vd

(b)

Figure 2.15: Decoupler leak flow: (a) system model and sign convention; (b) nonlinear resistance function.

Using equation (2.30) for losses through an orifice, the resistance on the leak flow is written as

P1 P2 = RL2 |QL | QL

(2.54)

It should be noted that the leak flow has not been modeled as high resistance flow through

the decoupler. Since the decoupler flow is integrated to attain the position, it is important to

preserve the memory of the decoupler plate by completely stopping the decoupler flow. There

are, however, two approaches to modeling the resistance to leak flow. The first is to consider

the lumped parameter RL2 as a constant and assume that during free decoupler motions negligible flow will pass through the leak orifice. The second approach is to develop a variable

resistance orifice which only exists when the decoupler is contacting the cage and the flow is in

the appropriate direction. In the following, the variable resistance function is developed.

Conditions for the nonlinear resistance function are

Vd Vd_ max ,

RL2,

,

Vd Vd_ max ,

RL2_f nc =

RL2,

Vd Vd_ max ,

,

Vd Vd_ max ,

Qd

Qd

Qd

Qd

>0

0

<0

0

(2.55)

The arctangent function is used again to provide the necessary continuous switching properties,

43

similar to the previous development of Ad_f nc . Starting with equation (2.50), the pressure differential is now replaced with the flow through the decoupler to give

2

C = arctan

2

Vd arctan( 5 Qd ) Vd_ max 6

(2.56)

where 5 and 6 are used to control the shape of the arctangent function, both with values of

1 105 . The value of C will be +1 for each case where Qd is in the positive direction and Vd

has exceeded Vd_ max or when Qd is in the negative direction and Vd is less than Vd_ max . All

other combinations of Qd and Vd will yield a C value of 1. From the nonlinear conditions of

RL2_f nc in equation (2.55), when C is +1 the desired value of RL2_f nc is RL2 , and when C is 1

the value of RL2_f nc should be . A function that provides this relationship is

RL2_f nc = RL2 (1 + ( 7 7 C))

(2.57)

where the value of 7 is the multiplicative factor that numerically approaches , and is 1 103

in this model. The complete nonlinear function for RL2_f nc using (2.56) and (2.57) is

RL2_f nc

2

2

= RL2 1 + 7 7 arctan

Vd arctan( 5 Qd ) Vd_ max 6

(2.58)

P1 P2 = RL2_f nc |QL | QL

(2.59)

Although not every decoupler design will exhibit leak flow, this particular development

will be used to model the mount studied in Chapters 5 and 6.

The final portion of this chapter is focused on developing a model that represents the hydraulic mount with a bell system.

44

Bell

Chamber

Bell

Upper

Chamber

Inertia

Track

Lower

Chamber

Decoupler

2.4

As illustrated in Figure 1.7b, the high frequency resonance of the decoupler creates an undesirable increase in the dynamic stiffness characteristics. A new design feature has been developed

that adds more flexibility to the high frequency results and alleviates the decoupler resonance

problem. The concept is to place an additional plate, or bell, within the upper chamber and

attached directly to the excitation point. The bell splits the upper chamber in two and creates

another flow passage, as illustrated in Figure 2.16. With the additional flow passage the system

takes on an another degree of freedom. In the following, a lumped parameter fluid system model

will be developed and the nonlinearity issues discussed. Also, a mechanical system representation is presented.

2.4.1

A lumped parameter model of the hydraulic mount shown in Figure 2.16 is illustrated in Figure 2.17. As seen in both Figures 2.16 and 2.17, the upper chamber is now divided in two with

the volume above the bell plate now referred to as the bell chamber and assigned the compliance

45

X(t)

Ap

Rb Ib

kr

br

Pb(t)

Ab

Qb(t)

Qi(t)

Cb

P1(t)

Qd(t)

C1

XT(t)

Ri Ii

Rd Id

FT(t)

P2(t)

C2

Am

Figure 2.17: Lumped parameter fluid system model including the bell.

parameter Cb . The chamber below the bell and above the decoupler and inertia track assembly

is still referred to as the upper chamber, keeping the same compliance parameter C1 . Above the

bell the effective pumping area is still denoted using parameter Ap . However, the area of the

mount at the bell is now captured with Am . The gap between the bell and mount wall can be

considered as an annular orifice with an area denoted by Ab . Under oscillating conditions, fluid

flow through area Ab is assigned lumped parameters Ib and Rb to capture the inertia and resistance, respectively. The inertia parameter represents a column of fluid communicating between

the bell chamber and the upper chamber. Although this inertia is quite low, it introduces an additional degree of freedom within the system which becomes apparent in the high frequency range.

The inertia track, decoupler and lower chamber receive the same lumped parameters used in the

original model.

Additional variables within the system include the pressure in the bell chamber Pb (t), and

the flow into the bell chamber Qb (t).

: -(Am-Ab-Ap)

: br

:Pb(t)

46

: Cb

: Rb

.

X(t)

: Qb(t)

: Ab

: 1/kr

Am-Ab

:P1(t)

: Ib

: C1

: Ii

: Id

: Qi(t)

: Qd(t)

: Ri

: Rd

: P2(t)

: C2

Figure 2.18: Bond graph model that includes the inertia track, decoupler and bell.

2.4.2

Using the lumped parameter fluid system model in Figure 2.17, the bond graph method is applied

to extract the system equations. Figure 2.18 represents a bond graph model for a hydraulic mount

with the bell system. Displacement excitation at the top of the mount is represented by a source

Three transformers are used to communicate between the linear motion and

of flow variable X.

displacements of fluid within the system. Each volumetric compliance parameter is connected

to the pressure nodes of the bell, upper and lower chamber; while, each flow variable resides

between the corresponding chamber pressures. Parameters and variables associated with the

bond graph model follow the same notation as in Figure 2.17. Since all experimental studies are

conducted with a fixed base test fixture, the motion of the base XT is set to zero.

2.4.3

47

System Equations

From the bond graph model, the continuity equations for each of the chambers are

Cb P b = Qb (Am Ab Ap )X

(2.60)

C1 P1 = (Am Ab )X Qb Qd Qi

(2.61)

C2 P2 = Qd + Qi

(2.62)

P1 Pb = Ib Q b + Rb (Qb + Ab X)

(2.63)

P1 P2 = Id Q d + Rd Qd

(2.64)

P1 P2 = Ii Q i + Ri Qi

(2.65)

These equations capture the internal dynamics of the hydraulic mount with a bell in the upper chamber, however they do not include the output force equation. The transmitted force to

the mount base is determined using the same technique as used for equation (2.6). As previously

determined, the transmitted force equation developed from the high frequency linear model captures all the desired characteristics and can be extended to the low frequency and nonlinear model

easily. Using Figure 2.17 and the linear form in (2.6), the transmitted linear force equation for

the free decoupler case is

FT = kr X + br X + (Am Ad )(P1 P2 ) + Am P2 + Ad Rd Qd (Am Ap )Pb

(2.66)

The additional term in (2.66) (Am Ap )Pb , represents the force exerted on the base from the bell

chamber pressure by ignoring the bell gap Ab . Equation (2.66) can be enhanced to a nonlinear

form with (2.53).

The set of equations (2.60) to (2.65) are of the same form as the linear equations first

presented for the standard hydraulic mount. This model can be handled similarly to the previous

48

cases, either by breaking the equations into two linear equations for low and high frequency

excitations, or by incorporating the nonlinear decoupler model.

2.5

The mechanical system model that includes the bell is developed using techniques similar to

thoes presented in the developments of the mechanical model for a standard hydraulic mount.

Since the mechanical system must be developed using linear system equations, the equations (2.60)

to (2.65) are considered for high frequency excitations only. The high frequency linear model

including just the decoupler and bell dynamics and becomes

Cb P b = Qb (Am Ab Ap )X

(2.67)

C1 P1 = (Am Ab )X Qb Qd

(2.68)

C2 P2 = Qd

(2.69)

P1 Pb = Ib Q b + Rb (Qb + Ab X)

(2.70)

P1 P2 = Id Q d + Rd Qd

(2.71)

Substituting equations (2.67) to (2.69) into the derivative of (2.70) and (2.71), gives the two

coupled liquid column system equations

1

1

1

Ib Qb + Rb Qb +

+

Qb +

Qd =

C1 Cb

C1

Am Ab Am Ap Ab

+

X Rb Ab X

C1

Cb

1

1

1

Am Ab

d + Rd Q d +

Id Q

Qd +

+

Qd =

X

C1

C1 C2

C1

(2.72)

Decoupler and bell flow variables are related to the velocity of the lumped fluid, by

Qd = Ad X d

Qb = Ab X b

(2.73)

49

where Ad and Ab are the areas of the decoupler and bell orifices. Also, the fluid inertia parameters are related to an equivalent lumped mass, and the resistance is converted to a linear damping

parameter

md

;

A2d

bd

=

;

A2d

Id =

Rd

mb

A2b

bb

Rb = 2

Ab

(2.74)

Ib =

The parameters md and mb represent the mass of the decoupler and bell, respectively. Also, bd

and bb are the linear damping parameters. Next, the compliance variables are transformed from

volumetric stiffness to a linear spring stiffness using,

Cb =

A2m

;

kb

C1 =

A2m

;

k1

C2 =

A2m

k2

(2.75)

where kb , k1 and k2 are the linear spring values representing each volumetric chamber stiffness.

Applying the relations developed in (2.73-2.75) to equation (2.72), gives

Ab

Ab

Ad

Ab

Ab

Ap

k1

1

X

Xb

Xd + kb

1

X

Xb

X =

Am

Am

Am

Am

Am

Am

Am

mb Xb + bb Xb X

Ab

Ab

Ab

Ad

Ad

k1

1

X

Xb

Xd k2

Xd =

(2.76)

Am

Am

Am

Am

Am

md Xd + bd Xd

Ad

To aid the development of a mechanical system that characterizes the equations in (2.76),

the mechanical lever system in Figure 2.19 is considered.

50

The equation for the displacement at Xc is

Ab

Ab

Xc = 1

X

Xb

Am

Am

(2.77)

and the lever is assumed to be weightless, thus the moment balance about X gives

Fc Ab = Fb Am

(2.78)

Applying the lever in Figure 2.19 and equations (2.77) and (2.78) to the moment balance

equations in (2.76), the mechanical system model shown in Figure 2.20 is formed.

To validate the model, the transmitted force to the base is developed and compared to

equation (2.66). Figure 2.21 illustrates the five points of force transmission to the base, which

are calculated as

FT _1 = kr X + br X

Ab

1

X

FT _2 = k1

Am

FT _3 = bd X d

(Am Ab )

FT _4 =

kb

1

Am

Ad

FT _5 = k2

Xd

Am

Ab

Ad

Ad

d

Xb

Xd k2

Xd bd X d md X

Am

Am

Am

Ab

Am

Ab

Ap

Xb

X

Am

Am

(2.79)

51

These equations are transformed to the fluid system form using (2.73) to (2.75) to give:

FT _1 = kr X + br X

FT _2 = (Am Ad )(P1 P2 )

FT _3 = Ad Rd Qd

(2.80)

FT _4 = (Am Ap )Pb

FT _5 = Am P2

Adding the transmitted force terms in (2.80) gives a resultant force term of

which is equivalent to (2.66), thus validating the mechanical system model for high frequency

excitations.

2.6

52

The bell system only introduces moderate additions to the nonlinear system equations previously developed. The inertia track and decoupler remain unchanged with respect to nonlinear

modeling. However, with the additional bell system, three chamber compliance values are now

required. New developments are not necessary for both the bell and upper chamber compliances, since equations (2.24) and (2.25) can be used to represent both. Depending on the level

of knowledge of each compliance, either a linear or nonlinear approach can be taken.

The bell flow resistance contributes to the nonlinear mount characteristics and is modeled

with the same approach used for the free decoupler. Starting with the flow resistance for steady

flow through an orifice (2.30), two nonlinear resistance parameters Rb1 and Rb2 are used to

characterize oscillating flow resistance. The nonlinear momentum equation for the bell becomes

+ Rb2 Qb + Ab X (Qb + Ab X)

P1 Pb = Ib Q b + Rb1 (Qb + Ab X)

(2.81)

The nonlinear transmitted force equation follows the same form as equation (2.53), however it receives an additional term representing the bell chamber influence. Using equation (2.66)

the nonlinear transmitted force becomes

(2.82)

where Ad_f nc is the decoupler area switching function (2.52). Equations (2.81) and (2.82) are

used in the linear model presented in (2.61) to (2.66) along with the nonlinear developments in

Section 2.3, to establish the complete hydraulic mount model that includes a bell. This complete

model is assembled in the following summary.

2.7 Summary

2.7

53

Summary

All aspects pertaining to the modeling of the hydraulic mount have been presented in this chapter. A linear model was developed first, and is representative of the cornerstone modeling work

presented by Singh et al. [30]. Bond graphs were used to develop system equations that are

comparable to the results documented in Lee et al. [15] and other publications [5, 3, 18]. Mechanical models developed from the linear systems help to illustrate the operating principles of

hydraulic mounts. These models are similar to the mechanical system representations first developed by Flower [5]. However, this work has mathematically validated the mechanical system

models using the transmitted force.

The development of a nonlinear system model represents advancements to hydraulic mount

modeling. Many nonlinear influences have been accounted for, including the effects of the upper

chamber compliance, inertia track resistance and decoupler switching effects. A summary of the

complete nonlinear system follows. The continuity equations are

C1 P1 = Ap X Qi Qd QL + C1 R1 Ap X Qi Qd

C2 P2 = Qi + Qd + QL

where the volumetric compliance parameters C1 (wdr , VT , Fp ), R1 (wdr , VT , Fp ) and C2 (Fp ) are

dependent on the frequency of excitation wdr , volume amplitude into the upper chamber VT and

the preload force Fp . The momentum equations are

P1 P2 = Ii Q i + (Ri1 + Ri2 |Qi |) Qi

P1 P2 = RL2_f nc |QL | QL

where

RL2_f nc

2

2

= RL2 1 + 7 7 arctan

Vd arctan( 5 Qd ) Vd_ max 6

2.7 Summary

54

FT = kr X + br X + (Ap Ad_f nc )(P1 P2 ) + Ap P2

+(Ad Ad_holes ) (Rd1 + Rd2 |Qd |) Qd

where

Ad_f nc

1

= Ad

arctan

2

2

Vd arctan ( 3 P ) Vd_ max 4

Fp

and the stiffness and damping parameters follow nonlinear behavior captured with kr wdr , X,

Fp ).

and br (wdr , X,

The nonlinear decoupler switching has been approached in several publications [11, 13, 18]

with only moderate success. These nonlinear decoupler models have modeled the switching

conditions with the pressure differential across the decoupler and the volume of fluid through

the decoupler. Although the published models represent the low frequency response, the high

frequency results are incorrect. The nonlinear decoupler model presented in this work operates

using the flow across the decoupler in combination with the volume. This model is effective

over the complete frequency spectrum of interest and will be validated in Chapter 6.

The final portion of this chapter has focused on a model that represents the hydraulic mount

with a bell system. Since, at the time of writing, no published material regarding this type of

system was recognized by the author, this portion of the modeling cannot be compared to other

models.

A lumped parameter model has been established by associating an inertia with the annular column of fluid between the bell plate and mount wall. A mechanical system model was

developed to represent a linear high frequency mount, including the bell and decoupler fluid column inertia. Finally, the nonlinear properties that are imposed by the bell system have been

considered.

2.7 Summary

55

The complete system model that includes the inertia track, decoupler and bell is summarized. The continuity equations are

C1 P1 = (Am Ab )X Qb Qd Qi QL

C2 P2 = Qi + Qd + QL

with the nonlinear volumetric parameters Cb (wdr , VT , Fp ), Rcb (wdr , VT , Fp ) and C2 (Fp ). The

bulge damping effect of the new upper chamber is neglected, since the compliance is generally

significantly lower than in the bell chamber.

The momentum equations are

P1 Pb

P1 P2 = RL2_f nc |QL | QL

where,

RL2_f nc

2

2

= RL2 1 + 7 7 arctan

Vd arctan( 5 Qd ) Vd_ max 6

FT = kr X + br X + (Am Ad_f nc )(P1 P2 ) + Am P2

+(Ad Ad_holes ) (Rd1 + Rd2 |Qd |) Qd (Am Ap )Pb

where,

Ad_f nc

1

= Ad

arctan

2

2

Vd arctan ( 3 P ) Vd_ max 4

2.7 Summary

56

Fp and br wdr , X,

Fp also follow the nonThe stiffness and damping parameters kr wdr , X,

linear behavior captured with

The complete model simulation techniques and validation with experimental system tests

are presented in Chapter 6; while, the following chapter provides a physical explanation of how

these models function.

Chapter

This chapter provides physical insight into the hydraulic mount dynamics using the mathematical models developed previously. To begin, the linear system models from Chapter 2 are used

to explain how liquid column resonance influences the output characteristics. The fluid column

resonance is developed in equation form and used to demonstrate the influence of parameters

within the linear model. The physical operation of the nonlinear decoupler model is also investigated using time and frequency domain simulations. Finally, the high frequency linear bell

model is used to explain how the addition of a bell influences the system characteristics. This

linear model is also used to investigate how system parameters influence output.

3.1

The linear models developed in Section 2.1.3 are used here to explain how internal liquid column resonance influences mount characteristics.

C1 P1 = Ap X Qi

C2 P2 = Qi

P1 P2 = Ii Q i + Ri Qi

57

58

FT = kr X + br X + Ap (P1 P2 ) + Ap P2

the behavior in the low frequency range is considered.

create a frequency domain equation for the cross point dynamic stiffness. Using the steady-state

sinusoidal solution detailed in Appendix B equation (B.13), the equation becomes

A2p

C2 A2p j

Ft

K (wdr j) =

= kr + br wdr j +

+

2

C2 C1 C12 )

X

With the transmitted force FT and input excitation X, the system response is characterized using

the cross-point dynamic stiffness K

FT

K = = |K (wdr j)|

= FT X = ]K (wdr j)

(3.1)

(3.2)

are the steady-state amplitudes of the transmitted force and excitation, respecwhere FT and X

tively.

To physically explain the internal dynamics, the model is simulated using the estimated linear parameters in Table 3.11 . Figure 3.1 shows the dynamic stiffness and phase characteristics

of the model using these estimated parameters. The simulation results and the characteristics

measured and plotted in Figure 1.7a, both indicate the effects of an internal liquid column resonance.

Table 3.1: Linear model parameters.

kr

br

Ap

Ad

C1

=

=

=

=

=

225 N/mm

0.1 103 kg/s

2500 mm2

660 mm2

3.0 104 mm5 /N

C2

Ii

Ri

Id

Rd

=

=

=

=

=

3.8 106 kg/mm4

10.5 105 kg/s-mm4

7.5 108 kg/mm4

11.7 106 kg/s-mm4

Units selected are those most commonly used in the automotive industry.

59

Phase (Degrees)

Frequency (Hz)

Phase (Degrees)

Volume (mm )

Figure 3.1: Dynamic stiffness and phase response of the linear low frequency model.

in-phase

out-of-phase

Frequency (Hz)

Magnitude

Phase

Pressure (Mpa)

Phase (Degrees)

(a)

Frequency (Hz)

Magnitude

Phase

(b)

Figure 3.2: Low frequency linear model response of: (a) volume passing through inertia track;

and (b) pressure in upper chamber.

60

As mentioned in Section 1.2, the liquid column resonance is indicative of the single DOF

(degree-of-freedom) fluid column bound by two compliant chambers. The resonant system is

best illustrated by observing the volume flow through the inertia track, shown in Figure 3.2a. The

volume response resembles a steady-state frequency domain curve of a single DOF system, with

a natural frequency of approximately 15 Hz. Before the system reaches resonance, the inertia

track fluid volume moves relatively in-phase with the excitation. At near-zero frequencies this

volume amplitude is equivalent to the product of excitation amplitude and effective pumping

area. The volume amplitude peaks at a phase of 90 , corresponding to the damped natural

frequency. At frequencies greater than the natural frequency, the volume amplitude diminishes,

moving relatively out-of-phase with respect to the excitation. This effect is sometimes referred

to as flow shut off. The two highlighted regions in Figure 3.2a are used to indicate areas where

significant flow is moving relatively in-phase and out-of-phase with the excitation. Flow through

the inertia track also changes the internal pressure differential, and influences overall hydraulic

mount characteristics.

The upper chamber pressure, illustrated in Figure 3.2b, dominates the hydraulic mount

system response. For frequencies around 10 Hz, the upper chamber pressure amplitude remains

relatively small and out-of-phase. As the pressure increases in magnitude and moves out-ofphase with the excitation, it decreases the dynamic stiffness magnitude and increases the phase

angle. This point is often referred to as the decoupler notch. Between 10 and 15 Hz the

phase angle peaks as increasing pressure begins to push the stiffness back in phase. From 15

to 20 Hz the pressure peaks, since a large amount of fluid is forced into the upper chamber by

the opposing fluid column motion and the excitation pumping motion. As a result, the dynamic

stiffness peaks with a reduced phase angle. Above 20 Hz, the flow magnitude diminishes and

the pressure decreases to a level comparable to a blocked inertia track.

61

An equation for the single DOF system, set up by the inertia and surrounding compliant

chambers, is developed by substituting (2.11) and (2.12) into the derivative of (2.13) to form

i + Ri Q i + (1/C1 + 1/C2 )Qi = Ap X

Ii Q

C1

(3.3)

i + 2wn Q i + w2 Qi = fo X

Q

n

where the undamped natural frequency of the liquid column is given by

s

(1/C1 + 1/C2 )

wn = 2fn =

Ii

(3.4)

(3.5)

=

Ri

2Ii wn

(3.6)

Ap

C1 Ii

(3.7)

fo =

From equation (3.5), the natural frequency is shown to be dependent only on the upper and

lower chamber compliances and the fluid column inertia. Figure 3.3 illustrates how the natural

frequency corresponds to the peak fluid column motion.

The high frequency linear model characteristics are examined using the equations (2.7)

to (2.10) developed in Section 2.1.3. These equations represent the low amplitude model that

assumes no inertia track

C1 P1 = Ap X Qd

C2 P2 = Qd

P1 P2 = Id Q d + Rd Qd

FT = kr X + br X + (Ap Ad )(P1 P2 ) + Ap P2 + Ad Rd Qd

62

Volume (mm3)

Phase (Degrees)

Phase (Degrees)

natural frequency fn

Frequency (Hz)

Stiffness Response

Volume (mm3)

Figure 3.3: Low frequency linear model response, showing the stiffness response and liquid

column motion.

Phase (Degrees)

Phase (Degrees)

natural frequency fn

Frequency (Hz)

Stiffness Response

Fluid Column Motion

Figure 3.4: High frequency linear model response, showing the stiffness response and liquid

column motion.

63

The cross point dynamic stiffness is simulated using the values in Table 3.1, and is shown in

Figure 3.4. Since equations (2.7) to (2.10) are of the same form previously examined, the same

internal liquid column resonance behavior results. The high frequency response demonstrates a

liquid column resonance of the decoupler corresponding to a natural frequency of 106 Hz.

From equations (2.7) to (2.9) the natural frequency of the decoupler is determined as:

s

(1/C1 + 1/C2 )

wn = 2fn =

(3.8)

Id

3.2

Since both linear models have similar dynamic properties, a general model is used to investigate

the influence of parameter changes on the system characteristics.

follows the same form as the two linear models previously investigated, however it is developed

using a generic inertia I and resistance R for the fluid column.

C1 P1 = Ap X Q

C2 P2 = Q

P1 P2 = I Q + RQ

(3.9)

(3.10)

(3.11)

FT = kr X + br X + Ap (P1 P2 ) + Ap P2

(3.12)

To simplify the model, this transmitted force is equivalent to equation (2.14), or (2.10) with Ad

set to zero.

Qualitative observations are now made regarding the impact of the compliance parameters, inertia and resistance. Also, the effective pumping area and linear stiffness and damping

parameters are investigated.

64

Phase (Degrees)

Frequency (Hz)

C1

C1 +20%

C1 +40%

Figure 3.5: Linear model response to changes in the upper chamber compliance parameter.

3.2.1

Compliance

Since the volumetric compliance of the lower chamber is several orders of magnitude higher

than the upper chamber, this parameter has little impact on the overall characteristics. This configuration is referred to as a single-load-bearing chamber mount, or SLBC [6]. If the lower

chamber compliance has a significant magnitude it will influence the dynamic stiffness prior to

resonance and will also act to influence the natural frequency through equation (3.5). In Figure 3.5 the influence of the upper chamber compliance parameter C1 is illustrated. An increase

in the upper chamber compliance decreases the liquid column resonant frequency, according

to equation (3.5). Beyond resonance, the increased compliance decreases the upper chamber

pressure and reduces the dynamic stiffness. The decreased natural frequency will increase the

nondimensional fluid column damping parameter through equation 3.6, thus reducing the peak

phase angle.

65

Phase (Degrees)

Frequency (Hz)

I

I +20%

I +40%

Figure 3.6: Linear model response to changes in the fluid inertia parameter.

3.2.2

Inertia

The inertia parameter I influences the liquid column resonant frequency, as shown in Figure 3.6.

From equation (3.5), an increase in fluid column inertia will decrease the resulting natural frequency. Also, the additional inertia will increase the amplitude of liquid column motion around

the natural frequency, thereby increasing pressure fluctuations and resulting in higher peaks in

the magnitude and phase characteristics. This point is further reinforced by observing that the

nondimensional damping parameter in equation (3.6) decreases with increased fluid column inertia.

66

Phase (Degrees)

Frequency (Hz)

R

R +20%

R +40%

Figure 3.7: Linear model response to changes in the fluid resistance parameter.

3.2.3

Resistance

The natural frequency of the liquid column is not altered by the resistance parameter R. Instead,

the resistance is responsible for changing the shape of the liquid column resonance through the

damping equation (3.6). Figure 3.7 illustrates how increased resistance decreases the peaks on

the dynamic stiffness curve.

during resonance, thus reducing pressure peaks and smoothing the total system response.

67

Phase (Degrees)

Frequency (Hz)

Ap

Ap +5%

Ap +10%

Figure 3.8: Linear model response for changes in the effective pumping area parameter.

3.2.4

Figure 3.8 illustrates the impact of the effective pumping area parameter Ap on the system characteristics.

The linear model does not fully capture its significance, since the effective area

parameter is mainly utilized in developing the mounts switching properties. However, it is important to illustrate how the effective area impacts the linear system. As illustrated in Figure 3.8,

and described by equation (3.5) to (3.7), the parameter Ap only influences the driving force on

the fluid column. Increasing the effective area raises the driving function (3.7) and results in

larger fluid column motion. Greater fluid column motion increases internal damping and upper

chamber pressures, resulting in increased peak phase angle and dynamic stiffness.

68

Phase (Degrees)

Frequency (Hz)

kr

kr +20%

kr +40%

Figure 3.9: Linear model response to changes in the linear stiffness parameter.

3.2.5

The stiffness kr and damping br parameters only influence the characteristics through the transmitted force equation. Since the damping inherent in rubber is relatively small (experimentally

established), the br parameter does not have a significant impact on the system characteristics.

An increase in the real stiffness parameter kr shifts the dynamic stiffness curve as illustrated in

Figure 3.9. Since kr only applies to the real stiffness, a reduction in phase angle results.

3.3

Of the nonlinear developments in Section 2.3, the decoupler switching is the most dominant.

This section examines the model response to confirm functionality and considers the impact of

the main nonlinear decoupler parameter. Frequency domain characteristics are used to explain

the decoupler operation over the complete range of excitations.

69

To validate the decoupler switching model operation, the linear model is enhanced with

the nonlinear decoupler developments of Section 2.3. These nonlinear equations include the

decoupler momentum, leak flow through the closed decoupler and the transmitted force. All

other nonlinear parameter developments are omitted for this case. The equations become

C1 P 1 = Ap X Qi Qd

(3.13)

C2 P 2 = Qi + Qd

(3.14)

P1 P2 = Ii Q i + Ri Qi

(3.15)

P1 P2 = RL2_f nc |QL | QL

(3.16)

(3.17)

(3.18)

where Ad_f nc (2.52) and RL2_f nc (2.58) are dependent on the nonlinear functions of decoupler

area and resistance, respectively. Developed to be

2

2

Vd arctan( 5 Qd ) Vd_ max 6

RL2_f nc = RL2 1 + 7 7 arctan

2

Ad

arctan

Vd arctan ( 3 P ) Vd_ max 4

Ad_f nc =

Due to the nonlinear nature of equations (3.13) to (3.18), a frequency domain equation

form cannot be derived. To obtain frequency response characteristics for this model, the nonlinear equations are simulated in the time domain by solving the set of simultaneous differential

equations, then converted to the frequency domain.

The equations are first rewritten as a set of first order differential equations in the form

y = F (t, y) and solved over a time interval [to , tf ], given initial values yo (to ). The solution

method uses numerical differentiation and a new technique for stiff differential problems [14].

Using MATLAB [19], the set of equations are solved for a particular excitation condition and

a Fourier transform is applied to the steady-state response to achieve the frequency data at the

70

kr

br

Ap

Ad

C1

C2

=

=

=

=

=

=

225 N/mm

0.1 103 kg/s

2500 mm2

660 mm2

3.0 104 mm5 /N

2.6 106 mm5 /N

Ri

Ii

Rd

Id

Rd3

RL2

=

=

=

=

=

=

3.8 106 kg/mm4

11.7 106 kg/s-mm4

7.5 108 kg/mm4

31.8 mm1

4.15 106 kg/mm7

maximum amplitude. For each excitation condition the procedure is repeated to determine the

complete frequency response curve.

Further details of the nonlinear simulation technique are covered in Chapter 6, where the

complete nonlinear model is simulated and validated with experimental data.

3.3.1

The decoupler switching characteristics are investigated for low frequency large, amplitude excitations by simulating the model with parameters given in Table 3.2. Excitations having peakto-peak amplitude levels of 1 mm and 2 mm are applied at 6 Hz to observe the decoupler volume, inertia track volume and upper chamber pressure. As Figure 3.10 indicates, the decoupler

equations impose the desired nonlinear mount characteristics. At approximately 350 mm3 the

decoupler plate contacts the cage, stopping flow and pushing fluid through the inertia track, as

illustrated in Figure 3.10b. These time domain plots also indicate the amplitude sensitivity inherent in the model. As the excitation amplitude decreases, less fluid is pushed into the upper

chamber, which reduces decoupler cage contact time. Since the decoupler is free a greater portion

of time, the upper chamber pressure and inertia track motion are both reduced. This characteristic is also observed in the frequency domain system response of Figure 3.11; dynamic stiffness is

reduced after resonance due to the reduction in upper chamber pressure. General characteristics

of the nonlinear system under large amplitude low frequency excitations are also comparable to

the linear model with only an inertia track.

71

Decoupler

Volume (mm3)

Inertia Track

3

Volume (mm )

(a)

Upper Chamber

Pressure (MPa)

(b)

Time (s)

(c)

2mm P-P

1mm P-P

Figure 3.10: Time domain simulation of nonlinear decoupler model: (a) decoupler volume; (b)

inertia track volume; (c) upper chamber pressure.

72

Phase (Degrees)

Frequency (Hz)

2 mm P-P

1 mm P-P

3.3.2

The same continuous nonlinear model is now simulated with the parameters in Table 3.2 over

0 to 250 Hz, illustrated in Figure 3.12. For an excitation amplitude of 0.05 mm the nonlinear decoupler model demonstrates characteristics corresponding to the linear model response in

Figure 3.4. Excitation levels of 0.1 mm and 0.2 mm also demonstrate the amplitude-sensitive

characteristics in the high frequency range.

pler fluid column motion enters the nonlinear region around resonance, shown in Figure 3.13.

Since the decoupler accommodates approximately 300 mm3 , the volume of fluid displaced into

the upper chamber is limited and the peak upper chamber pressure is diminished. These effects

influence the system characteristics by reducing the peak dynamic stiffness and phase angle.

The frequency response characteristics in Figures 3.12 and 3.13 introduce an interesting

phenomena. Since the new hydraulic mount model is now a nonlinear set of differential equations, the frequency response is multi-valued. The resulting steady-state solution now has more

73

Phase (Degrees)

Frequency (Hz)

0.05 mm P-P

0.075 mm P-P

0.1 mm P-P

Phase (Degrees)

Volume

3

Magnitude (mm )

Figure 3.12: Dynamic stiffness characteristics of the nonlinear model under high frequency

excitations.

Frequency (Hz)

0.05 mm P-P

0.075 mm P-P

0.1 mm P-P

Figure 3.13: Decoupler motion in the nonlinear model under high frequency excitations.

74

than one possible value for a given excitation condition. This system response is not only dependent on the amplitude and frequency of excitation, but is also dependent on the initial conditions.

As cited in Inman [7], a unique characteristic of nonlinear systems is a bend in the frequency

response, produced by unstable regions. To demonstrate that this unstable region exists within

the nonlinear decoupler model, frequency response curves have been calculated by sweeping

the excitations. Using this technique, the initial conditions for a particular frequency become

the last conditions of the previous response. The impact of the initial conditions can be illustrated by comparing the frequency response for forward swept excitations and backward swept

excitations.

Figure 3.14 shows how the amplitude and phase of the decoupler motion are dependent

on excitation sweep direction, furthermore the unstable region becomes evident. However, the

impact of this phenomenon is somewhat diluted in the system dynamic characteristics. As Figure 3.15 illustrates, the dynamic stiffness is not greatly influenced by initial conditions, however,

the phase does show a small region of instability.

To compare this simulation approach with the technique used for experimental evaluation,

both Figures 3.15 and 3.14 include the frequency response obtained using a mean dwell method.

This method brings the system to rest for a period of time between excitations, to allow all

internal dynamics to settle and resetting all initial conditions to zero. From the results, it is

concluded that an acceptable frequency response is obtained by setting the initial conditions to

zero for each excitation condition. Since the mount testing procedure also uses a mean dwell sine

sweep process, all further nonlinear simulations will be conducted by setting initial conditions to

zero for each excitation.

75

Phase (Degrees)

Volume

Magnitude (mm3)

Frequency (Hz)

Forward Sweep

Backward Sweep

Mean Dwell

Phase (Degrees)

Figure 3.14: Decoupler motion indicating region on instability and dependence on sine sweep

conditions.

Frequency (Hz)

Forward Sweep

Backward Sweep

Mean Dwell

Figure 3.15: Dynamic stiffness and phase indicating impact of sine sweep conditions.

3.3.3

76

The switching characteristics of the nonlinear decoupler model are controlled with the parameter

Rd3 . This parameter limits the fluid volume through the decoupler via equation (3.17), reflecting the decoupler cage height. To verify the perceived impact of the decoupler parameter Rd3

on the system response, the model is simulated with an excitation of 6 Hz and 2 mm P-P. Using the parameters in Table 3.2, the time domain decoupler volume Vd is plotted in Figure 3.16.

Simulation results verify a direct relation between Rd3 and the volume of fluid passing through

the decoupler. Increasing the decoupler parameter limits the volume of fluid through the decoupler, indicative of a decreased cage height. Details of this parameter-to-geometry relation are

presented in Chapter 7.

To complete the discussion on the operation of the nonlinear decoupler model, the parameter Vd_ max is considered. This parameter represents the maximum fluid volume amplitude

allowed through the decoupler. Since this characteristic is also related to the decoupler cage

height it will be shown that Vd_ max is a function of the decoupler parameter Rd3 . The relation-

Decoupler

Volume (mm3)

Time (s)

Rd3

Rd3 +30%

Figure 3.16: Simulation of decoupler volume indicating the response to changes in the decoupler parameter.

3.4

77

A bell plate is incorporated into the hydraulic mount to alleviate the undesired high frequency

resonance of the decoupler liquid column. The bell model, introduced in Section 2.4, assigns

inertia and flow parameters to the annular column of fluid between the bell plate and mount wall.

This new bell fluid column places an additional DOF within the mount. Experimental studies

show that the bell is only effective at high frequencies. This can be explained by noticing that

the bell inertia is several orders of magnitude less than that of the inertia track. Thus, the low

frequency high amplitude characteristics are not significantly altered. This section, therefore,

only examines the high frequency response of the linear bell and decoupler model.

The high frequency linear model that includes the decoupler and bell dynamics (with Qi

set to zero), is developed using equations (2.60) to (2.66). The linear equations become

Cb P b = Qb (Am Ab Ap )X

(3.19)

C1 P1 = (Am Ab )X Qb Qd

(3.20)

C2 P2 = Qd

(3.21)

P1 Pb = Ib Q b + Rb (Qb + Ab X)

(3.22)

P1 P2 = Id Q d + Rd Qd

(3.23)

(3.24)

The two degrees of freedom now present within the system, stem from the momentum

equations (3.22) and (3.23). To simulate the frequency response of this system, equations (3.19)

to (3.23) are first manipulated into the matrix form in Appendix B equation (B.21)

Ib 0

0 Id

b

Q

d

Q

Rb 0

0 Rd

Q b

Q d

"

Am Ab

C1

1

C1

+

1

C1

1

Cb

1

C1

Am Ap Ab

Cb

Am Ab

C1

1

C1

+

#

1

C2

Qb

Qd

Rb Ab

0

78

This system of equations is then converted to the frequency domain assuming the steady-state

solutions

b ejwdr t

Qb = Q

d ejwdr t

Qd = Q

(3.25)

jwdr t

X = Xe

The frequency domain responses of Qd and Qb are obtained by solving the characteristic equation and isolating the variables in matrix form. With the flow equations, the pressure

response in each chamber is developed and used to calculate the transmitted force via equation (3.24). The cross point dynamic stiffness K and phase angle , are determined using the

same approach presented in Section 3.1. Appendix B contains details on the equation manipulation for simulation.

Table 3.3: Parameters used in linear bell model.

kr

br

Ap

Ab

Ad

Am

=

=

=

=

=

=

225 N/mm

0.1 103 kg/s

2500 mm2

600 mm2

660 mm2

4900 mm2

Cb

C1

C2

Rd

Id

Rb

Ib

=

=

=

=

=

=

=

700 mm5 /N

2.6 106 mm5 /N

11.7 106 kg/s-mm4

7.5 108 kg/mm4

2.8 107 kg/s-mm4

2.0 108 kg/mm4

Using estimated parameters, given in Table 3.3, the high frequency dynamic stiffness and

phase responses are simulated and shown in Figure 3.17. Response characteristics illustrate how

the additional bell plate alleviates the increased stiffness from decoupler resonance. To explain

physically the internal dynamics generating these unique characteristics, the frequency response

of both bell and decoupler fluid columns are considered along with the chamber pressures in

Figures 3.18 and 3.19.

The two coupled differential equations in (B.21) represent a two degree of freedom system that contains eigenvalues, or natural frequencies, solved through the characteristic equation.

79

Phase (Degrees)

Frequency (Hz)

Figure 3.17: Dynamic stiffness and phase characteristics from the simulated linear model that

includes the bell and decoupler.

Each natural frequency corresponds to mode shapes pertaining to the internal fluid column variables Qb and Qd , flow through the bell and decoupler respectively. Following the same approach

as in Section 3.1, the fluid column motions are correlated to the internal pressures which directly

influence the systems dynamic stiffness characteristics.

modes of vibration, frequency response curves indicate characteristics not present in the standard hydraulic mount configuration.

Table 3.4: Frequency and mode shape vectors of high frequency bell model.

Natural Frequency:

Mode 1

94 Hz

Mode 2

1525 Hz

Mode Shape:

(1, 1.02)T

(1, 0.26)T

From the eigenvalue solution in Table 3.4, the system undergoes a first resonant frequency

at 94 Hz. For illustrative purposes the phase in Figure 3.18 has been adjusted 180 to be aligned

with respect to the excitation; the original coordinate direction of Qb is opposite that of Qd .

80

Volume (mm3)

Phase (Degrees)

Phase

Frequency (Hz)

Magnitude

(b)

Phase

Volume (mm3)

Phase (Degrees)

Frequency (Hz)

Magnitude

(a)

Pressure (Mpa)

Phase (Degrees)

Figure 3.18: High frequency linear model response of: (a) decoupler fluid column motion; and

(b) bell fluid column motion.

Phase

Frequency (Hz)

Magnitude

(b)

Phase

Pressure (Mpa)

Phase (Degrees)

Frequency (Hz)

Magnitude

(a)

Figure 3.19: High frequency linear model response of: (a) upper chamber pressure; and (b)

bell chamber pressure.

81

During this mode of excitation the decoupler and bell oscillate at 90 phase to the excitation.

Physically, this mode of vibration represents both decoupler and bell inertia values moving inphase, thus the first natural frequency has decreased to 94 Hz, from the previous value of 106 Hz

with just a decoupler. This trend follows from equation (3.8), where increases in inertia decrease

the natural frequency. The resulting dynamic stiffness and phase characteristics represent those

of a standard hydraulic mount up to approximately 100 Hz.

At frequencies between 100 and 150 Hz, decoupler motion decreases drastically at a phase

of 180 to the excitation. Pressure in the upper chamber also decreases rapidly in this frequency range, as the bell fluid column motion moves 180 to the excitation and accommodates

fluid displaced into the upper chamber.

stiffness also undergoes the same trend, producing another low point or notch in the curve in

Figure 3.17. From this explanation, it will be shown in Chapter 6 that even without a decoupler

the mount exhibits the same decrease in stiffness.

The frequency response of the decoupler volume indicates an anti-resonance occurring at

156 Hz (see Figure 3.18a). Antiresonance is a local effect in MDOF systems, occurring only

in certain response curves depending on the modal coordinates. This is explained further in Maia

et al. [17]. Physically, the antiresonance implies the fluid column is grounded at this frequency

or its real motion is zero. Also characteristic of the antiresonance is the 180 phase shift, which

is shown in Figure 3.18a.

At frequencies beyond 156 Hz the decoupler motion begins to increase, now in-phase with

the excitation. The upper chamber pressure amplitude also increases at 180 to the excitation.

As the pressure increases it drives the dynamic stiffness up and forces the response out of phase.

This trend continues through the frequency range of interest.

From Table 3.4, the second resonant frequency in the system occurs at 1525 Hz and has

a mode shape with the decoupler and bell fluid column 180 out of phase. This fluid collision

82

in the upper chamber results in high pressures and high dynamic stiffness. The design of a hydraulic mount with a bell system will, therefore, not remove the internal liquid column resonance

of the decoupler, but will give the engineer greater ability to tune high frequency characteristics.

3.5

The additional bell dynamics create a system capable of achieving better isolation characteristics

in the high frequency range. However, the multiple degrees of freedom within the system also increase the complexity. Since the characteristic equation for the system must be solved to obtain

the eigenvalues, the natural frequencies cannot be written explicitly in terms of the parameters.

The natural frequencies and mode shape vectors in Table 3.4 are approximated using the mass

and stiffness matrices from equation (B.21) and solving the eigenvalue problem, given known

parameters. To better understand how the parameters influence the system response, the antiresonance of the decoupler will be considered along with simulations illustrating the influence of

parameter changes.

As previously discussed, the antiresonance frequency of the decoupler fluid column is associated with the notch point in the dynamic stiffness curve. Theoretically, when the equation

representing the decoupler frequency response in Figure 3.18a is zero, the decoupler fluid column is at antiresonance. From equations (3.19) to (3.23), the frequency domain equation for the

decoupler flow is determined in Appendix B equation (B.27), as

d =

Q

(Am Ab )wdr

C1

1

Cb

Am Ap Ab

(Am Ab )Cb

2

Ib wdr

+ Rb

Characteristic Equation

Rb Ab

Am Ab

wdr j

where the Characteristic Equation is the determinant of the inertia, resistance and compliance

matrix transformed into the frequency domain.

zero at antiresonance, it also corresponds to the point of zero flow. Therefore, the frequency

of antiresonance occurs when the real component of the numerator in equation (B.27) is zero.

Solving for the antiresonance driving frequency wdr yields

v

u

u 1 1 Am Ap Ab

t Cb

Am Ab

wdr = wa = 2fa =

Ib

83

(3.26)

where wa and fa denote the notch or antiresonance frequency in radians per second and Hertz,

respectively. Substituting the parameters from Table 3.3 into equation (3.26) yields a frequency

of 156 Hz, corresponding to the notch frequency previously observed.

highlights that antiresonance is not a function of decoupler inertia or middle chamber compliance, but rather a function of bell chamber compliance, fluid column inertia and effective area

parameters within the mount.

The influence of equation (3.26) will be confirmed by observing the effect on model simulation results to changes in compliance, inertia, resistance and area parameters. Stiffness and

damping parameters kr and br have been omitted from this analysis since they were shown previously to have minimal impact on the system response.

3.5.1

Compliance

Two of the three compliance values will be investigated. The lower chamber compliance is omitted since it has an insignificant impact on the system characteristics, leaving the bell and upper

chamber to be studied. Figure 3.20 illustrates changes in system characteristics to variations in

the upper chamber compliance C1 parameter. Since the compliance of this chamber is several

orders of magnitude smaller than the bell compliance, these simulations require a large percentage increase in C1 to illustrate the influence. From the calculated resonant frequencies tabulated

in Table 3.5 an increase in the upper chamber compliance will reduce both natural frequencies,

however the second mode indicates a larger decrease. Dynamic stiffness and phase responses

indicate the forward shift in liquid column resonances and a stationary antiresonance frequency,

corresponding to equation (3.26). Both the tabulated data and the frequency response confirm

that the notch location is not influenced by the upper chamber compliance.

84

Table 3.5: Frequency modifications to changes in the upper chamber compliance parameter.

Parameter

C1

C1 +1000%

C1 +3000%

Mode 1

94 Hz

88 Hz

78 Hz

Mode 2

1525 Hz

514 Hz

336 Hz

Antiresonance

156 Hz

156 Hz

156 Hz

The bell chamber compliance parameter Cb is also considered. As Table 3.6 indicates, the

bell compliance influences the first resonant frequency, while the high frequency mode remains

relatively unchanged. Increases in the bell compliance parameter decrease the first mode of vibration and shifts the notch frequency, shown in Figure 3.21. Changes in antiresonance location

are again indicated by equation (3.26). Note that changes in system characteristics at the first

resonance are similar to results found in the single DOF system (see Figure 3.5).

Table 3.6: Frequency modifications to changes in the bell chamber compliance parameter.

Parameter

Cb

Cb +40%

Cb +80%

Mode 1

94 Hz

80 Hz

71 Hz

Mode 2

1525 Hz

1521 Hz

1520 Hz

Antiresonance

156 Hz

132 Hz

116 Hz

85

Phase (Degrees)

Frequency (Hz)

C1

C1 +1000%

C1 +3000%

Phase (Degrees)

Figure 3.20: Dynamic stiffness and phase response to changes in the upper chamber compliance parameter.

Frequency (Hz)

Cb

Cb +40%

Cb +80%

Figure 3.21: Dynamic stiffness and phase response to changes in the bell chamber compliance

parameter.

3.5.2

86

Inertia

Figure 3.22 and Table 3.7 highlight the impact of the decoupler fluid column inertia parameter Id

on the system characteristics. An increase in decoupler inertia decreases both natural frequencies within the system. Figure 3.22 demonstrates that dynamic stiffness characteristics at the

first resonance change shape similarly to the single inertia system illustrated in Figure 3.6. Furthermore, antiresonance within the system is not influenced by changes to the decoupler inertia.

Table 3.7: Frequency modifications resulting from changes in the decoupler fluid column inertia parameter.

Parameter

Id

Id +40%

Id +80%

Mode 1

94 Hz

62 Hz

74 Hz

Mode 2

1525 Hz

1480 Hz

1454 Hz

Antiresonance

156 Hz

156 Hz

156 Hz

The impact of the bell fluid column inertia Ib is illustrated in Figure 3.23 and Table 3.8.

The plot shows little change in the first natural frequency to increases in the bell column inertia, however the second mode frequency decreases significantly. Also, the response shows a decreased antiresonance frequency as the notch shifts to the right, corresponding to equation (3.26).

Since the bell fluid column inertia is significantly lower than that of the decoupler, the first mode

of vibration does not undergo much change as the bell inertia is altered.

Table 3.8: Frequency modifications resulting from changes in the bell fluid column inertia

parameter.

Parameter

Ib

Ib +40%

Ib +80%

Mode 1

94 Hz

91 Hz

87 Hz

Mode 2

1525 Hz

1341 Hz

1226 Hz

Antiresonance

156 Hz

132 Hz

117 Hz

87

Phase (Degrees)

Frequency (Hz)

Id

Id +40%

Id +80%

Phase (Degrees)

Figure 3.22: Dynamic stiffness and phase response to changes in the decoupler fluid column

inertia parameter.

Frequency (Hz)

Ib

Ib +40%

Ib +80%

Figure 3.23: Dynamic stiffness and phase response to changes in the bell fluid column inertia

parameter.

88

Phase (Degrees)

Frequency (Hz)

Rd

Rd +40%

Rd +80%

Figure 3.24: Dynamic stiffness and phase response to changes in the decoupler resistance

parameter.

3.5.3

Resistance

Figure 3.24 includes the frequency response curves of dynamic stiffness and shows the impact of

changes in the decoupler resistance parameter Rd . Since the resistance is independent of internal

stiffness and damping, changes in resistance have no influence on the resonance or antiresonance

frequencies within the system. Instead, decoupler resistance is shown to decrease the peaks on

the first mode, similar to changes in the single DOF system illustrated in Figure 3.7.

Changes to the bell resistance parameter Rb have no significant impact on the system characteristics. The bell resistance parameter reflects the resistance to fluid flow between the mount

wall and the bell plate, and generally has values that are several orders of magnitude less than

the decoupler resistance.

89

Phase (Degrees)

Frequency (Hz)

Am

Am +10%

Am +20%

Figure 3.25: Dynamic stiffness and phase response to changes in the mount area parameter.

3.5.4

Area Parameters

Finally, the influence of internal area parameters are considered. First it should be highlighted

that the natural frequencies for both modes of vibration are not influenced by any area parameters,

since the eigenvalue problem only uses the inertia and compliance matrices of equation (B.21).

Table 3.9 and Figure 3.25 provide evidence that Am primarily changes the antiresonance frequency, or notch of the dynamic stiffness characteristics. These results are again supported by

equation (3.26).

Table 3.9: Frequency modifications resulting from changes in the mount area parameter.

Parameter

Am

Am +10%

Am +20%

Mode 1

94 Hz

94 Hz

94 Hz

Mode 2

1525 Hz

1525 Hz

1525 Hz

Antiresonance

156 Hz

148 Hz

141 Hz

90

The effective pumping area Ap also influences the notch frequency. Figure 3.26 shows how

the changes in stiffness response at the first resonance are similar to the single DOF system in

Figure 3.8. As the effective area increases, more fluid is pushed through the system, increasing

damping and pressure levels during resonance. The antiresonance or notch frequency is also

altered through equation (3.26) and is tabulated in Table 3.10.

Table 3.10: Frequency modifications resulting from changes in the effective pumping area

parameter.

Parameter

Ap

Ap +10%

Ap +20%

Mode 1

94 Hz

94 Hz

94 Hz

Mode 2

1525 Hz

1525 Hz

1525 Hz

Antiresonance

156 Hz

164 Hz

171 Hz

Subtle changes in the notch frequency result from changes in the bell area Ab . Figure 3.27

gives an indication of simulation results for increased bell area parameter. Table 3.11 reinforces

that the antiresonance increases with increases in Ab .

Table 3.11: Frequency modifications resulting from changes in bell area parameter.

Parameter

Ab

Ab +40%

Ab +80%

Mode 1

94 Hz

94 Hz

94 Hz

Mode 2

1525 Hz

1525 Hz

1525 Hz

Antiresonance

156 Hz

161 Hz

166 Hz

A cautionary note should be made about the Am and Ab parameters. In most cases, modifying these parameters will alter the bell inertia and resistance parameters, yielding slightly different results than shown here. Chapter 7 will present some preliminary suggestions and results

on the relation between Am , Ab and the bell fluid column parameters.

91

Phase (Degrees)

Frequency (Hz)

Ap

Ap +10%

Ap +20%

Phase (Degrees)

Figure 3.26: Dynamic stiffness and phase response to changes in the effective pumping area.

Frequency (Hz)

Ab

Ab +40%

Ab +80%

Figure 3.27: Dynamic stiffness and phase response to changes in the bell area parameter.

3.6 Summary

3.6

92

Summary

The main focus of this chapter was to provide a better understanding of the models developed

in Chapter 2.

By investigating the simulation results, it was first concluded that the models

demonstrate similar trends to experimental data. After the linear models were deemed feasible,

they were used to physically explain the dynamics within the mount. Numerical studies were

conducted using the linear models to highlight the impact of system parameters on the overall

characteristics.

Preliminary nonlinear model simulations showed how this model captures the desired response across the complete frequency and amplitude spectrum of interest. Insight into the decoupler switching parameters was also documented.

Finally, the bell model has been shown to contain additional properties for tuning high

frequency characteristics within the mount. Dynamic characteristics of the bell and decoupler

system were physically explained using a linear, high frequency model. By applying modal

analysis techniques the high frequency notch, indicative of a bell system, has been associated

with an antiresonance of the decoupler fluid column. After physically explaining the system

response, the influence of parameters was investigated.

All work conducted in this chapter gives a reasonable understanding of the mount operation, however there is still a lingering question. What are the actual parameter values within the

system that pertain to a particular hydraulic mount? The next half of this thesis is dedicated to

resolving this issue.

Chapter

Experimental Design

Up to this point, the hydraulic mount has been modeled and simulated to demonstrate the internal

dynamics and influence of system parameters. The thesis now moves into quantifying these

system parameters by introducing an experimental apparatus designed to isolate and identify the

hydraulic mount system parameters.

The first portion of this chapter is dedicated to establishing the goals and approach to the

experimental process. The available resources and identification techniques are developed into

experimental design criteria, which are then used for the final design.

Instrumentation and

4.1

One of the primary purposes of this research is to accelerate hydraulic mount design. System

models and simulations presented thus far, give a good indication of the function and influence

of several parameters; however, the models lack some relation to actual hydraulic mount components. As a result, this work begins to focus on measuring parameters from existing hydraulic

mount components.

The approach is conceptually simple. Since all models are based on lumped parameter

assumptions, it is possible to break apart the mount and identify parameters in isolation. For

93

94

completely characterizes this component through the parameters Ii , Ri1 and Ri2 . If the inertia

track is isolated and a controlled flow input is applied, the pressure response from this equation

can be used to determine the unknown parameters.

Component isolation and parameter identification is the first step to bridging the gap between the mathematical models and physical hydraulic mount systems (see Figure 4.1).

- Isolate components

- Identify parameters

Nonlinear Lumped

Parameter Models

Physical System

Response

4.2

Design Criteria

perspective, the experimental system had to be capable of providing oscillatory input conditions

to match those found within the mount. Also, the data acquisition system had to be capable of

reading small deviations in pressure and transient flow in the time domain. Available equipment

limitations required that the flow be selected as the controlled input to the apparatus. The system

had to operate with internal pressures of 0 to 50 psi (gauge) with the capability to safely reach

100 psi. These criteria were established from hydraulic mount pressure measurements cited in

Kim [10].

95

When establishing design criteria it was determined that an existing MTS servo-controlled

hydraulic rate machine (mount testing system) could provide high frequency controlled excitation to the apparatus. This approach reduced the cost and complexity of the system while

utilizing high precision, computer interfaced equipment.

The physical construction of the apparatus had to accommodate all components while interfacing both mechanically and electrically with the rate machine. Attention was also placed

on a rigid apparatus design to reduce any undesired dynamics during testing.

4.3

Final Design

The experimental apparatus in Figure 4.2 (photographs in Figure 4.3 and 4.4) uses a hydraulic

cylinder

1 , driven via the servo-controlled actuator. With the base of the cylinder clamped

to the machine base, linear motion of the actuator displaces fluid into a two chamber vessel.

Displacement control on the actuator becomes volume control by multiplying the fixed piston

area by the linear actuator motion. While keeping the piston top open to atmosphere, actuator

2 into the identification vessel

3.

motion pushes fluid through a 12 -in diameter union joint

The 5 12 -in inside diameter vessel is constructed with 12 -in thick cast acrylic, contained between

aluminum connectors.

4 , clamped between vessel

sections, to fix the various components in line with the controlled flow. High frequency pressure

sensors

5 are used to collect output data on each side of the holding plate, capturing the pressure

differential. Complete assembly drawings and a material list of the final design are compiled in

Appendix C. Details on the setup of specific component tests is presented in Chapter 5.

96

Excitations from

servo-controlled actuator

5

4

97

98

4.4 Instrumentation

Servo-controlled

Actuator

Experimental

Test Rig

Pressure

Transducers

99

LVDT

Controller

Command

A/C Conditioner

D/C Conditioners

Data Acquisition

Workstation

4.4

Instrumentation

The apparatus design made full use of the existing equipment so that the only new instrumentation required were two pressure transducers. Two Sensotec subminiature Model S pressure

transducers were selected for their high natural frequency, enabling accurate readings for transient pressure measurements. Since pressure and force transducers are electrically similar, the

two pressure sensors were interfaced directly to the MTS controller through existing force transducer ports and internal signal conditioners. The pressure transducers were directly conditioned

and sampled using the existing real-time high speed, data acquisition system. In some instances

the pressure signals were even used for feedback on the actuator. This arrangement enabled

precise pressure measurements to be acquired with actuator position readings, necessary for

measuring transient behavior. Figure 4.5 illustrates the instrumentation configuration.

4.5 Calibration

4.5

100

Calibration

To interface the pressure transducers with the MTS controller software, a sensor calibration file

was created using the transducer calibration procedure [21]. Pressure sensor calibration was

conducted using a certified measurement tool, consisting of a pneumatic pump and digital pressure gauge, illustrated in Figure 4.6. Pressure signals were calibrated for 0 to 50 psi and the

units were left in Imperial units until post data collection processing.

Pressure

Transducer

air lines

Signal to controller

Pneumatic

Hand Pump

Digital Pressure

Guage

Curiously, the pressure signals were calibrated such that a positive pressure within the

vessel reads as negative. Therefore, the maximum pressure of 50 psi was actually read as 50 psi

by the MTS system and in the data. This configuration used the same convention for the existing

actuator calibration, measuring a downward motion as negative, and was necessary for pressure

feedback control within the vessel.

In addition, the 0 to 50 psi calibrated pressure range does not accommodate possible vacuum conditions within the vessel. To ensure that locations within the test apparatus never droped

below atmospheric pressure, all tests were conducted under elevated pressure conditions. Details

of the mean operating pressure will be stated as the individual tests are documented.

4.5 Calibration

4.5.1

101

The actuator displacement measurement Xact from the LVDT are multiplied by the hydraulic

piston area Apiston to establish the volume of fluid that passes into the vessel, or volume excitation

Vexcit ,

Vexcit = Xact Apiston

(4.1)

where the piston area is 5350 mm2 . The flow is then calculated as the time derivative of the

volume excitation

Qexcit = V excit

(4.2)

Although the test apparatus was designed to be as rigid as possible, a portion of fluid from

the excitation volume Vexcit will be accommodated by the volumetric expansion of the vessel. To

determine the actual fluid volume that passes through the holding plate Vactual it is first necessary

to identify the compliance associated with the lower vessel Crig , as illustrated in Figure 4.7. The

damping associated with this expansion is assumed negligible.

Xact

compliance Crig

Ps2

Vactual

Apiston

Ps1

Crig

Vexcit

Figure 4.7: Illustration of the volumetric compliance associated with the vessel lower chamber.

To determine the volumetric compliance, the flow across the holding plate is first blocked

with a rigid plate. Then, using the pressure sensor for feedback, the lower vessel chamber is

4.5 Calibration

102

-40

-35

-30

-25

-20

-15

-10

-5

-0.1

-0.2

-0.3

-0.4

-0.5

Measurements

Trendline

pressurized using the actuator driven piston. Data from the pressure sensor Ps1 and actuator

displacement Xact are plotted in Figure 4.8.

2

Xact = 1.2 104 Ps1

+ 0.0175Ps1

(4.3)

Multiplying by the piston area (4.1) gives the volume displaced in the lower chamber. Using this

volume, the slope at some instantaneous pressure Pinst in the lower vessel chamber represents

the compliance

Crig =

Apiston Xact

= Apiston 2.4 104 Pinst + 0.0175

Ps1

(4.4)

To determine the actual volume of fluid passing through the holding plate during normal

operation, the time domain measurements of pressure Ps1 are used to modify actuator displacement Xact . First, the differential data series Xact and Ps1 are calculated using,

k+1

k

Xact = Xact

Xact

(4.5)

103

Vexcit = Apiston Xact

(4.6)

The instantaneous pressure in the lower chamber is then approximated as the average between

the k and k + 1 measurements

k+1

k

Pinst = Ps1

+ Ps1

/2

(4.7)

Using equation (4.4) the lower chamber compliance is then calculated and used with Ps1 to

modify the volume displaced by the actuator. The result is the actual volume change, written as

Vactual = Vexcit Crig Ps1

(4.8)

With the actual change in volume passing through the holding plate (4.8), the cumulative volume

Vactual is determined using

k+1

k

Vactual

= Vactual

+ Vactual

(4.9)

From the preceding development, the volumetric compensation procedure can only be conducted in the time domain. To obtain frequency domain measurements, the actuator position and

chamber pressures data must be first collected in the time domain. At the end of each frequency

and amplitude test condition, both input excitation and output pressures are sampled and written

to a file. The time domain data is processed using equations (4.1) to (4.9) to calculate the actual

volume, or flow, passing into the system. Frequency analysis is then conducted on the output

pressures and calculated input. For all analysis procedures presented in the following chapter,

the input to the isolated system will be compensated to account for the vessel compliance.

4.6

Apparatus Capabilities

pressure inside the identification vessel should not exceed 50 psi gauge. For safety, the vessel

104

pressure should not exceed the maximum pressure of 100 psi. The pressure transducers have

been calibrated from 0 to 50 psi, therefore readings are only accurate within these bounds. For

all tests a safety shut-off, or hydraulic interlock, was set to initiate once a pressure of 50 psi was

exceeded. It is recommended that this precaution be taken for all future tests conducted with the

apparatus.

Accuracy is also limited in the frequency domain. Inherent flexibility in the apparatus

induced a dominant mode of vibration at approximately 70 Hz, which altered the chamber pressure readings and skewed measurements. However, this limitation did not impede experimental

studies since all tests were conducted below 70 Hz.

specific test configurations and system identification techniques used to extract hydraulic mount

parameters.

Chapter

Parameter Identification

The experimental apparatus presented in Chapter 4 is now used to identify the model parameters

established in Chapter 2. Throughout this chapter each main component within the hydraulic

mount system is studied in isolation.

system identification techniques are documented. Enough information is provided to enable the

extension of this work to encompass other mount designs. The design under investigation in this

chapter is the Cooper-Standard mount shown in Figure 5.1.

5.1

During the lumped parameter assignments in Section 2.1.1, the upper compliance was given four

parameters kr , br , Ap and C1 representing the linear stiffness, damping, effective pumping area

and volumetric compliance, respectively. To enhance these characteristics, the model developments in Section 2.3.1 assigned a resistance parameter to the bulge damping effect and made

all four stiffness and compliance parameters nonlinear. The following section presents the approach used to sort through each upper compliance parameter, beginning with the real stiffness

and damping characteristics.

105

106

Upper

Compliance

Decoupler

Inertia

Track

Lower

Compliance

5.1.1

The nonlinear stiffness kr and damping br parameters are defined to be a function of excitation

driving frequency wdr , and preload force Fp . From (2.22), these parameters are

amplitude X,

written as,

Fp

kr wdr , X,

Fp

br wdr , X,

Since these parameters are only dependent on the conditions applied to the whole mount system,

the internal fluid system dynamics have no influence on the stiffness and damping parameters.

The approach to investigating these parameters is to remove the fluid from the hydraulic mount

and conduct tests on the upper compliance, effectively in isolation. Test data for the stiffness

and damping parameters are collected using the same configuration and procedure used for a

hydraulic mount, presented in Section 1.2.

107

Measurements

The static force-displacement curve is measured and plotted in Figure 5.2. This curve

illustrates the hysteresis inherent in rubber, which gives most of the materials damping properties. The dynamic characteristics are measured using steady-state sinusoidal excitations with

the preload force held constant at 1700 N, reflecting in-vehicle conditions, and summarized in

Table 5.1.

Figure 5.3 shows the measured nonlinear stiffness and damping parameters found using

the test conditions in Table 5.1. Since these parameters have been established over the excitation

amplitude and frequency ranges of interest and are independent of internal fluid dynamics, they

were included in a look-up table for future system model simulations.

Table 5.1: Test conditions used for dynamic characterization, at 1700 N preload.

P-P Amplitude (mm)

1

2

0.1

0.2

0.3

2 to 40

2 to 40

10 to 250

10 to 250

10 to 250

2

2

10

10

10

kr (N/mm)

108

br (N s/mm)

br (N s/mm)

kr (N/mm)

Frequency (Hz)

Frequency (Hz)

(a)

1 mm P-P Amp

(b)

2 mm P-P Amp

Figure 5.3: Stiffness and damping properties measured at various excitation conditions: (a)

low frequency high amplitude; (b) high frequency low amplitude.

5.1.2

The nonlinear model for the effective pumping area Ap (Fp ) is dependent on the preload force.

As the mount is displaced under the preload force, the rubber deforms so that the effective piston

area changes. To investigate this relationship, the experimental apparatus shown in Figure 5.4

was used to isolate the upper compliance. By completely filling the vessel with fluid below the

compliance, ensuring no air pockets are present, the changes in displacement at the mount top

pushed fluid into the lower vessel chamber. Using pressure sensor feedback, the vessel pressure

was maintained at 5 psi.

control repositioned the actuator to accommodate the fluid displaced. Measurements of mount

displacement and actuator position were tabulated and used to determine the effective area as a

function of mount displacement, shown in Figure 5.5. To obtain the appropriate effective area

parameter using the preload force, the static stiffness curve in Figure 5.2 was first used to obtain

the preload displacement. For the mount under investigation, the effective pumping area was

found to be 3065 mm2 at 1700 N preload.

109

Mount

Displacement

Xact

Upper

Compliance

Ps1

Pressure

Feedback

Actuator

Position

Fluid

Filled

Chamber

Ap (mm2)

Figure 5.4: Test configuration for measuring the effective area and volumetric compliance

parameters.

Measurements

5.1.3

110

Once the stiffness, damping and effective pumping area parameters are established, the volumetric compliance parameters can be investigated. Two approaches are used to extract the C1

and R1 parameters. Tests are first conducted on the chamber in isolation. Then a simplified

hydraulic mount is tested and used in conjunction with the system model to establish the same

parameters.

Using the test configuration illustrated in Figure 5.4, the upper compliance is displaced by

the amount corresponding to the 1700 N preload. A known volume of fluid is then forced into

the upper chamber and pressure readings are collected. The static curve of volume displacement

versus pressure is shown in Figure 5.6. These measurements illustrate the bulge damping effect

from the rubber, which is modeled by introducing the resistance parameter R1 . Previous experimental studies documented by Kim and Singh [12] limit the investigation of compliance to static

analysis and neglect the effects of hysteresis. However, the present work includes experimental

studies on the dynamic volumetric compliance and bulge damping.

Dynamic characteristics of the volumetric compliance and bulge damping parameters,

C1 (wdr , VT , Fp ) and R1 (wdr , VT , Fp ) are investigated by establishing a local linear response for

various driving frequencies wdr and volume amplitudes VT . This is achieved by applying sinusoidal actuator excitations and measuring lower chamber pressure response.

The excitations are applied under a controlled mean pressure of 15 psi, and steady-state

time domain data is acquired for actuator position Xact and pressure Ps1 . The actual volume of

fluid passing into the upper chamber Vactual is then computed using equations (4.1) to (4.9). The

parameters C1 and R1 are identified through the compliance model developed in Section 2.3.1,

equation (2.24),

C1 P1 = QT + C1 R1 Q T

(5.1)

111

Volume (mm3)

Pressure (MPa)

Measurements

Figure 5.6: Static curve of volume versus pressure in the upper chamber.

For the isolated experimental tests the variables QT and P1 are replaced by the measured

system input and output variables Qactual and Ps1 , respectively, where Qactual is the derivative of

Vactual . Integrating (5.1) with respect to time and assuming all initial conditions are zero, gives

Ps1 =

Vactual

+ R1 V actual

C1

(5.2)

To cast equation (5.2) into the frequency domain, the steady-state pressure and volume variables

take the form

Ps1 = Ps1 ejwdr t

(5.3)

where Ps1 and Vactual represent the steady-state amplitude and phase of pressure and volume,

respectively. The final frequency domain equation becomes,

Ps1

1

=

+ R1 wdr j

C1

Vactual

(5.4)

the real and imaginary components of equation (5.4). The parameters C1 and R1 are identified

using the measured frequency domain data and are illustrated in Figure 5.7.

A few points should be noted regarding this test configuration. First, the frequency range is

limited to under 70 Hz due to the limited capabilities of the test apparatus. Therefore, volumetric

112

compliance parameters are not identified at high frequencies. In addition, the test configuration

fixes the upper compliance so that the normal excitation point is stationary. Since this point is

in motion during excitations, the test setup does not exactly represent the physical system.

To address the above mentioned issues, the C1 and R1 parameters are also identified using

a special case hydraulic mount.

decoupler is used to extract the compliance parameters. The inertia track geometry generates

high inertia and damping properties which result in a liquid column resonance below 20 Hz.

Figure 5.8 illustrates that the inertia track has negligible influence on the system above 30 Hz.

Since the stiffness, damping and effective pumping area parameters are already identified, the

model response in the high frequency range is tuned by adjusting the C1 and R1 parameters

to match the dynamic stiffness characteristics of the special mount.

volumetric parameters are identified from 30 Hz to 250 Hz and illustrated in Figure 5.9.

The upper compliance parameters in (2.25) are dependent on the fluid volume amplitude

being pushed into the upper chamber. Since this volumetric expansion is a function of internal

fluid dynamics, the C1 and R1 parameters cannot be directly calculated at each excitation condition. To establish parameter values during the model simulations, an iterative procedure is

applied to update compliance parameters based on the steady-state volume response. Using an

initial guess of C1 and R1 , the steady-state simulation response is used to calculate volume flow

amplitude VT into the upper chamber. The compliance parameters are updated using a surface

fit to the data in Figure 5.9, then the simulation is reiterated. Complete details of the simulation

procedure are presented in Chapter 6.

113

C1 (mm /N)

R1 (Ns/mm5)

Frequ

ency (H

m

Volu

z)

e Am

ude

p lit

(m m

Frequency

(H

(a)

me

Volu

z)

de

litu

Amp

(m m

(b)

Phase (Degrees)

Phase (Degrees)

Dynamic

Stiffness (N/mm)

Dynamic

Stiffness (N/mm)

Figure 5.7: Upper compliance parameters identified at 15 psi mean pressure using the test

apparatus: (a) volumetric compliance; and (b) bulge damping.

Frequency (Hz)

Frequency (Hz)

(a)

1 mm P-P Amp

(b)

2 mm P-P Amp

Figure 5.8: Dynamic stiffness and phase response of the special mount, measured using the

test conditions in Table 5.1: (a) low frequency; and (b) high frequency response.

114

C1 (mm /N)

R1 (Ns/mm )

Frequ

ency (

Hz)

um

Vol

m

e (m

itud

l

p

m

eA

(a)

Frequency

(Hz)

me

Volu

(

ude

plit

Am

)

mm

(b)

Figure 5.9: Upper compliance parameters identified using the system model and test conditions

in Table 5.1: (a) volumetric compliance; and (b) bulge damping.

5.2

(SLBC) mount, since the lower chamber serves mainly as a fluid transfer area [6]. Essentially,

the lower chamber compliance acts as a diaphragm, contributing little to the volumetric stiffness

of the system. However, the assignment of a compliance parameter, C2 (Fp ) (2.26), to the lower

chamber is still carried out. This parameter is only dependent on the preload force, or the mean

volume of fluid in the lower chamber.

To isolate the C2 parameter, the test apparatus was used with the lower chamber fastened

across the holding plate, as illustrated in Figure 5.10. The static volume versus pressure curve

was measured and is plotted in Figure 5.11.

During model simulation the C2 parameter is calculated by first estimating the volume of

fluid displaced into the lower chamber under the preload force. This is done using the effective

pumping area curve in Figure 5.5 and the static stiffness curve in Figure 5.2.

The slope of

Figure 5.11 at this volume level is used to approximate the volumetric compliance of the lower

Lower

Compliance

115

Xact

Actuator

Position

Ps1

Pressure

Fluid

Filled

Chamber

Volume (mm3)

Figure 5.10: Test configuration for determining the volumetric compliance of the lower chamber.

Pressure (MPa)

Measurements

Figure 5.11: Static measurements of lower chamber volumetric expansion versus pressure.

116

chamber. It should be noted that some hydraulic mounts are filled after the preload force is

applied, thus the volume displaced is different than that mentioned here.

For the mount under investigation, the volume displaced was found to be 32944 mm3 at

the 1700 N preload, and the resulting lower chamber volumetric compliance was calculated to

be 2.6 106 mm5 /N.

5.3

Inertia track parameters are identified experimentally using the test apparatus configuration shown

in Figure 5.12. The decoupler and inertia track assembly is bolted to the holding plate within the

identification vessel. By blocking the decoupler, the fluid flow across the holding plate V actual

becomes the inertia track flow and transducer measurements Ps1 and Ps2 form the pressure differential in the momentum equation. To ensure that pressure levels within the vessel never exceed

the 0 to 50 psi range, the vessel is sealed with an air pocket and pressurized to a mean operating level of 20 psi. Using pressure feedback on Ps2 , the upper vessel chamber is controlled to

operate at the desired mean level during all test conditions.

The nonlinear inertia track model developed in Section 2.3 equation (2.29) is

P1 P2 = Ii Q i + (Ri1 + Ri2 |Qi |) Qi

where the parameters Ii , Ri1 and Ri2 are first investigated using frequency sweep data and a

simplified linear momentum equation. Sinusoidal excitations are applied to the actuator and

steady-state time domain data is acquired of the actuator position Xact and both chamber pressures Ps1 and Ps2 . Fluid volume passing through the inertia track Vactual is then calculated and

used with the pressure differential Ps1s2 = Ps1 Ps2 in the isolated system equation. The

isolated response follows a linearized form of equation (2.29),

Ps1s2 = Ii Vactual + Ri V actual

(5.5)

117

Cap to

seal vessel

X

Inertia

Track

Fluid

Fill

Level

Actuator

Position

Ps2

Ps1

Pressure

Sensors

Figure 5.12: Test configuration for isolating the inertia track paramters.

This equation is transformed to the frequency domain using the steady-state solutions,

Ps1s2 = Ps1,s2 ejwdr t

(5.6)

where Ps1,s2 and Vactual represent the complex pressure differential amplitude and volume amplitude, respectively. The linear parameters are then identified using

Ps1s2

2

= Ii wdr

+ Ri wdr j

Vactual

(5.7)

Measured frequency response data is calculated by applying the Fourier transform on steady-state

time domain data of Vactual and Ps1s2 . The Ii and Ri parameters are plotted in Figure 5.13

for several excitation conditions. The trends confirm that the fluid column inertia remains constant regardless of flow through the inertia track and suggest that the resistance parameters are

dependent on the flow, as indicated by the nonlinear momentum equation (2.29).

x 10

-6

x 10

118

-4

Resistance (kg/s-mm4)

Inertia (kg/mm4)

2 .5

1 .5

1 .5

0 .5

0 .5

0.5

1.5

2.5

3.5

0.5

1.5

2.5

x 10

3.5

4

5

(a)

x 10

(b)

Figure 5.13: Parameter trends with respect to inertia track flow, for: (a) inertia; and (b) resistance.

The resistance parameters in equation (2.29) are approximated using the trendline shown

in Figure 5.13b where Ri1 is the y intercept and Ri2 is the slope. For the inertia track under

investigation, the parameters are determined to be

Ii = 3.29 106 kg/mm4

Ri1 = 4.9 105 kg/s-mm4

Ri2 = 4.6 1010 kg/mm7

Although this approach is sufficient for determining the nonlinear inertia track parameters,

a more elegant method can be applied which significantly reduces testing time. To introduce this

method, equation (2.29) is rewritten in terms of the measured experimental variables.

Ps1s2 = Ii Q actual + (Ri1 + Ri2 |Qactual |) Qactual

(5.8)

where Qactual represents the time derivative of Vactual . For each time sample of pressure differential and flow, equation (5.8) defines the relationship between variables within the isolated

system. In theory, this suggests that three samples of measured data is enough to set up three

119

equations and solve each of the desired parameters. If a sequence of time domain data is acquired, the least squares method will determine the best estimate of system parameters. This

approach first requires the sequence of measurements be placed in matrix form, as follows

1

1

Qactual 1 Qactual 1 |Qactual | Qactual

Ps1s2

I

2

2

2

i

2 Ps1s2

Qactual Qactual

|Qactual | Qactual

=

(5.9)

Ri1

..

..

..

..

R

.

.

.

.

i2

n

n

Ps1s2

Qactual n Qactual n |Qactual | Qactual

By assigning each matrix to the notation,

1

Qactual

Ps1s2

2

2

Ps1s2

Q actual

; U =

Y =

.

..

..

.

n

n

Ps1s2

Qactual

Qactual

2

Qactual

..

.

Qactual

|Qactual | Qactual

2

|Qactual | Qactual

..

.

|Qactual | Qactual

1 T

= U T U

U Y

Ii

= Ri1

Ri2

(5.10)

where is the least squares estimate of the parameters in , (see Wilson [34] or Juang [9])

To apply this technique, a random perturbation is applied to the actual volume input as

illustrated in Figure 5.14a. The frequency spectrum of the excitation is measured and plotted in

Figure 5.14b, which indicates whether the perturbation is within the frequency range of interest.

The random excitation is applied for 10 seconds and sampled every 0.002 seconds. Data samples

are assembled into the form presented in (5.9) to obtain the inertia track parameters and the

estimated output illustrated in Figure 5.15. Parameter estimation using this technique identifies

Ii = 3.38 106 kg/mm4

Ri1 = 6.52 105 kg/s-mm4

Ri2 = 4.49 1013 kg/mm7

These results indicate that both the sine sweep and random least squares method can be applied

successfully to extract the inertia track parameters. Since the random perturbation method is

120

11

Flow (mm3/s)

Flow

6 2

Spectral Magnitude (mm /s )

x 10

3

2.5

2

1.5

1

0.5

0

10

15

20

25

30

Frequency (Hz)

Time (s)

(a)

(b)

Figure 5.14: Isolated inertia track excitation plots including: (a) time domain segment of random flow input; and (b) power spectral density of complete flow perturbation.

0.1

0.05

-0.05

-0.1

2.05

2.1

2.15

2.2

2.25

2.3

2.35

2.4

2.45

2.5

Time (s)

Measured

Figure 5.15: Time domain segment of measured pressure differential versus model output

using estimated parameters.

121

experimentally faster and computational less intensive than the sine sweep method, it is recommended for inertia track parameter identification.

5.4

Decoupler Parameters

The experimental test configuration used to identify the decoupler parameters is shown in Figure 5.16. To ensure that all flow forced across the holding plate passes through the decoupler,

the inertia track is blocked. The actual volume Vactual is used with the output pressure differential Ps1s2 to identify the system parameters through the momentum equation. Pressure levels

within the vessel are maintained at a mean operating level of 20 psi to ensure the transducer

limits are not exceeded.

The nonlinear decoupler model established in Section 2.3 equations (2.36) and (2.59) include a nonlinear momentum equation in parallel with a variable resistance orifice,

P1 P2 = RL2_f nc |QL | QL

respectively. Five parameters in equations (2.36) and (2.59) control the complete behavior of

the nonlinear decoupler model. The inertia Id and resistances Rd1 and Rd2 govern the decoupler

properties when operating in the uncoupled region, while Rd3 controls the decoupler switching. Also, the nonlinear leak flow resistance RL2_f nc , exists as RL2 when the decoupler is closed

and is controlled with the parameter Vd_ max .

equations (2.36) and (2.59) represent a complex five parameter identification task with one state

variable Vd . As a consequence, the following approach utilizes a two step identification procedure.

The first experimental test is a frequency sweep, similar to the inertia track test. If the

excitation volume amplitude is such that the decoupler is kept within the uncoupled region, the

122

Cap to

seal vessel

Xact

Fluid

Fill

Level

Decoupler

Actuator

Position

Ps2

Ps1

Pressure

Sensors

system response will follow equation (2.31)

P1 P2 = Id Q d + (Rd1 + Rd2 |Qd |) Qd

which is a reduced form of equation (2.36). Simplifying equation (2.31) into a linear form with

the experimentally measured variables gives,

Ps1s2 = Id Vactual + Rd V actual

(5.11)

This linear momentum equation is then converted to the frequency domain using the steady-state

expressions in (5.6).

Ps1s2

2

= Id wdr

+ Rd wdr j

Vactual

(5.12)

Equation (5.12) is used to identify the inertia and resistance parameter trends, as conducted

for the inertia track. However, the important distinction to this frequency domain test is the

dependance on volume amplitude. For all excitation conditions, equation (5.12) only applies if

the response is linear, or the decoupler does not contact its cage. For this reason, each steady-

123

Pressure (MPa)

Pressure (MPa)

Volume (mm )

Volume (mm5)

Time (s)

Time (s)

(a)

(b)

Figure 5.17: Steady-state time domain data of actual decoupler volume and pressure differential at 12 Hz, showing: (a) nonlinear cage contact; and (b) linear uncoupled response.

state response is first checked to ensure the pressure differential does not indicate decoupler cage

contact. Figure 5.17 illustrates the desired characteristics of steady-state response data.

The Fourier transform is applied to the steady-state data to establish the frequency response

characteristics and identify Id and Rd using equation (5.12). Identified parameters are plotted

over the decoupler flow amplitude in Figure 5.18.

The isolated response demonstrates that the decoupler inertia remains constant across various flow conditions, while the resistance increases with the decoupler flow; validating the nonlinear uncoupled equation (2.31). A reasonable approximation for parameters Rd1 and Rd2 is

established using the intersection and slope of the linear trendline. For the mount under investigation, the parameters are found to be

Rd1 = 4.58 106 kg/s-mm4

Rd2 = 4.25 1011 kg/mm7

The amplitude-sensitive behavior of the decoupler gives unreliable results using the least

squares method. Therefore, the sine sweep technique provides the best method to identifying

parameters pertaining to the uncoupled decoupler.

x 10

-8

x 10

124

-6

Resistance (kg/s-mm )

Inertia (kg/mm4)

7

6

4

3

Flow (mm3/s)

Flow (mm3/s)

x 10

(a)

8

4

x 10

(b)

Figure 5.18: Parameter trends with respect to decoupler flow, for: (a) fluid column inertia; and

(b) resistance.

The two remaining parameters Rd3 and RL2 combine with the decoupler state Vd , or unknown plate position Xd , to form a state and parameter identification problem. Since the switching characteristics under investigation are independent of excitation frequency, the second isolated test is conducted by applying low frequency random perturbations. Under these excitations the decoupler inertia has negligible impact on the isolated system response, therefore it is

removed from equations (2.36) and (2.59). Also, the nonlinear leak flow resistance function

RL2_f nc is reduced to the RL2 parameter, since the flow is controlled when isolated in the test

apparatus. The isolated switching model becomes,

Ps1,s2 =

(5.13)

(5.14)

where 1 and 2 are maintained at constant values of 1 1016 and 1 105 , respectively, as

established in Section 2.3. Also, recall from equation (2.33) that

Xd =

Vd

Ad

125

where the decoupler area Ad is measured from geometry of the mount under investigation.

In equations (5.13) and (5.14) the actual flow controlled across the decoupler Qactual now

represents the total volume accommodated via the decoupler and leak flow,

Qactual = QL + Qd

(5.15)

Also, the absolute value signs in equations (5.13) and (5.14) are resolved into algebraic equations

by noting the signs of Qd and QL are equivalent to that of the input Qactual . Using the arctangent

function presented in Section 2.3, the absolute values of decoupler and leak flow are transformed

according to:

2

Qd arctan(Qactual )

2

(Qactual Qd ) arctan(Qactual )

|Qactual Qd | =

|Qd | =

(5.16)

(5.17)

Using (5.15) to (5.17), the equations (5.13) and (5.14) are transformed to

2

Ps1s2 = Rd1 Qd + Rd2 Q2d arctan(Qactual ) + 1 eRd3 Xd arctan(Qactual 2 ) Qd

2

Ps1s2 = RL2 (Qactual Qd )2 arctan(Qactual )

(5.18)

(5.19)

Equations (5.18) and (5.19) are equated to form a function for decoupler flow Qd , by solving the

quadratic equation

2

0 =

(RL2 Rd2 ) arctan(Qactual ) Q2d

(5.20)

4

Rd3 Xd arctan(Qactual 2 )

Rd1 Qd + 1 e

+ RL2 Qactual arctan(Qactual ) Qd

2

+RL2 Q2actual arctan(Qactual )

simplified into a function of the state Vd , parameters Rd3 and RL2 , and input Qactual

Qd = f (Vd , Rd3 , RL2 , Qactual )

(5.21)

126

Using the decoupler flow (5.21) in equation (5.19) gives the system output equation, also written

in reduced form as

Ps1s2 = g(Vd , Rd3 , RL2 , Qactual )

(5.22)

Equations (5.21) and (5.22) form the state and parameter identification problem, where

the decoupler state has to be predicted and each parameter estimated, for each position in time.

Convergence issues arise when attempting to predict Vd and identify Rd3 and RL2 all in one

algorithm. Therefore, RL2 is calculated prior to predicting the state variable.

The leak flow resistance is first calculated using the measured pressure differential output,

as illustrated in Figure 5.19. The peak pressure differential measurements (circled locations on

Figure 5.19) represent the high resistance leak flow around the closed decoupler, occurring when

QL = Qactual . The RL2 parameter is then calculated using

RL2 =

Ps1s2

Q2

actual

(5.23)

where Ps1s2 and Qactual represent the measured data at the pressure differential peaks.

Once RL2 is estimated, the functions in (5.21) and (5.22) are reduced to just one unknown

parameter Rd3 , and the state variable Vd .

Qd = f (Vd , Rd3 , Qactual )

(5.24)

(5.25)

Qd,k =

Vd,k+1 Vd,k

T

(5.26)

where k represents the k th sample point of the data series, sampled at time intervals T . Using

(5.26) to transform equations (5.24) and (5.25) into discrete form gives,

Vd,k+1 = f (Vd,k , Rd3,k , Qactual,k , k)

(5.27)

(5.28)

127

Pressure (MPa)

Flow (mm5/s)

Time (s)

Figure 5.19: Time segment of the decoupler input flow and pressure differential, indicating

locations where leak resistance is identified.

Identification of both states and parameters presents a unique problem. First, the state

estimation problem formulation assumes knowledge of all system parameters. The system identification problem to determine parameters then assumes the availability of the variables or states

within the system. To work around this problem the parameter Rd3 is written as a state variable and appended to the original state (see Wilson [34]). Since Rd3 is constant, the new state

equation becomes

Rd3,k+1 = Rd3,k

(5.29)

this state is then added to Vd,k+1 to form the new state equation vector

k+1 =

X

Vd,k+1

Rd3,k+1

Rd3,k

(5.30)

k =

X

Vd,k

Rd3,k

(5.31)

128

k+1 = f (X

k , Qactual,k , k)

X

(5.32)

k , Qactual,k , k)

Ps1s2,k = g(X

(5.33)

The new state vector (5.31) transforms the system into a nonlinear state estimation problem, which is then solved using the Extended Kalman Filter

k+1 = f (X

k , Uk , k) + wk

X

k , Uk , k) + qk

Zk = g(X

(5.34)

(5.35)

where Uk and Zk represents the system input and output, respectively. The wk denotes disturbance noise on the states, while qk represents measurement noise on the output. The complete

Extended Kalman Filter algorithm used to identify state variables from equations (5.34) and

(5.35), is presented in Appendix D.

Random perturbations are applied to the isolated system for approximately 20 seconds and

sampled at time intervals T = 0.004 s. Input and output measurements are used to solve for

RL2 , then analyzed using the filter algorithm. The results in Figure 5.20 show how the procedure

converges on parameter estimates, while Figure 5.21 provides a closer look at the predicted state

motion and output accuracy. For the decoupler under investigation this identification procedure

obtained

RL2 = 9.21 109 kg/mm7

Rd3 = 14.5 mm1

The Vd_ max parameter, used as the switching parameter in the nonlinear transmitted force

(2.52) and leak flow (2.58) equations, is determined using the converged decoupler state estimate

Vd . As Figure 5.21b indicates, Vd_ max has an estimated value

Vd_ max = 750 mm2

0.06

5

Volume (mm )

2000

0.04

Pressure (MPa)

129

0.02

0

1000

0

-1000

-2000

-0.02

10

15

20

15

20

(b)

19.2

-0.04

Rd3

16.0

-0.06

12.7

-0.08

-0.1

9.5

0

10

15

3.2

20

10

Time (s)

Time (s)

(a)

(c)

Figure 5.20: Decoupler state and parameter identification: (a) pressure differential across decoupler; (b) decoupler volume state variable; and (c) decoupler switching parameter.

1000

Volume (mm )

0.05

Pressure (MPa)

0.04

0.03

0.02

500

0

-500

-1000

0.01

15

16

17

18

19

18

19

(b)

0

-0.0

16.0

Rd3

-0.02

12.7

-0.03

9.5

-0.04

15

16

17

18

Measured

19

15

16

17

Time (s)

Time (s)

Estimate

(c)

(a)

Figure 5.21: Close up of the decoupler state and parameter identification: (a) pressure differential across decoupler; (b) decoupler volume state variable; and (c) decoupler switching parameter.

5.5

130

Bell Parameters

The bell system is also experimentally studied using the same Cooper-Standard mount. A bell is

included in the upper chamber such that when the 1700 N preload is applied, the bell operates 5

to 10 mm from the decoupler plate, as illustrated in Figure 5.22. In the current Cooper-Standard

mount, the operating position of the bell is such that the bell chamber compliance is equivalent

to the previously measured upper chamber. To obtain full knowledge of all parameters, this

configuration only requires the identification of the bell fluid column inertia, resistance, and new

upper chamber compliance.

5.5.1

The mathematical model developed to represent the bell system in Section 2.4 assigns inertia

and resistance parameters to the annular column of fluid between the bell plate and mount wall.

These parameters are now experimentally measured using the test apparatus configuration illustrated in Figure 5.23. A section of the mount wall is clamped into the holding plate within the

apparatus. The bell plate is then fixed into position between the wall segment with a 12 -in wide

bracket. To control fluid flow through the bell, the bracket is designed so that it does not obstruct

flow around the bell plate. The fluid volume Vactual pushed across the holding plate becomes the

bell flow, while the pressures Ps1 and Ps2 are measured. As with the inertia track and decoupler

tests, the vessel is sealed with an air pocket and pressurized to a mean operating level of 20 psi.

The nonlinear bell momentum from equation (2.81) is

where P1 Pb is the pressure differential across the bell plate, equivalent to the measured differential in the isolated case. The variable X represents the bell plate motion in the system model,

and is zero for the isolated test configuration. Therefore, equation (2.81) can be simplified to a

131

Upper

Compliance

Bell Plate

Bell Position

With Preload

Decoupler

Inertia

Track

Lower

Compliance

Figure 5.22: Cross section of Cooper-Standard mount that includes a bell plate.

Cap to

seal vessel

wide

bracket

Segment of

mount wall

Fluid

Fill

Level

Xact

Actuator

Position

Bell Plate

Ps2

Ps1

Pressure

Sensors

132

Ps1 Ps2 = Ib Q actual + (Rb1 + Rb2 |Qactual |) Qactual

(5.36)

similar to the isolated inertia track and decoupler tests. The linear moment is converted to the

frequency domain, using the steady-state expressions in (5.6),

Ps1s2

2

= Ib wdr

+ Rb wdr j

Vactual

(5.37)

This expression is used to identify the inertia and resistance parameters from the Fourier

transforms of the steady-state measurements. The identified parameters are plotted over the bell

flow amplitude in Figure 5.24. These experimental results also demonstrate a constant fluid

column inertia over the complete flow range, while increases in resistance are evident with increased bell flow. The trends validate the nonlinear momentum equation (5.36), with reasonable

approximations for parameters Rb1 and Rb2 established using the intersection and slope of the

linear trend line. For the mount under investigation, the parameters were found to be

Ib = 2.27 108 kg/mm4

Rb1 = 2.33 108 kg/s/mm4

Rb2 = 2.97 1012 kg/mm7

5.5.2

Chamber Compliance

The position of the bell was selected such that the bell compliance, or chamber above the bell, is

the upper compliance already investigated in Section 5.1, leaving the middle chamber unknown.

5.6 Summary

x 10

133

-8

x 10

-6

1 .2

Resistance (kg/s-mm4)

Inertia (kg/mm4)

1 .5

0 .8

0 .6

0 .4

0 .5

0 .2

1.5

2.5

3.5

Flow (mm3/s)

1.5

5

x 10

(a)

2.5

Flow (mm3/s)

3.5

5

x 10

(b)

Figure 5.24: Parameter trends with respect to bell flow, for: (a) inertia; and (b) resistance.

As illustrated in Figure 5.22, the chamber between the bell plate and inertia track - decoupler

housing assembly, referred to now as the upper chamber, will contain negligible bulge damping

and very little volumetric compliance C1 .

To approximate C1 , it is assumed that the steel housing deflects under pressure and gives

the chamber some volumetric compliance. The inertia track - decoupler housing is modeled

as a 79-mm diameter and 1.8-mm thick steel disk, and finite element methods are applied to

approximate the plate compliance.

illustrates the constraints and distributed pressure load. The linear deflection in Figure 5.25b

is used to calculate the volume displaced under the pressure applied. Using this approach, the

volumetric compliance of the upper chamber is found to be

C1 = 680 mm5 /N

It should be noted that although this volumetric compliance value is in the order of one tenth

that of the bell, it cannot be assumed rigid. The stiffness between the bell and decoupler flow is

necessary to establish the coupled multi DOF system.

5.6 Summary

Pressure Load

134

3.

3.

3.

3.

3.

3.

3.

3.

3.

3.

F

3.

F

3.

F

3.

24

3.

24

3.

3.

24

3.

24

3.

24

3.

24

24

F

15

24

24

24

3.

24

15

15

15

24

15

15

15

24

15

15

15

15

24

15

15

15

15

24

1245

Constraints

(a)

(b)

Figure 5.25: Illustration of the quarter plate model used to approximate the chamber compliance, showing: (a) boundary constraints and 3 MPa evenly distributed pressure load; (b) plate

deflection in linear elastic region.

5.6

Summary

This chapter has provided a systematic look at measuring each of the mount components in

isolation. Each of the upper chamber compliance parameters were identified through one of

three tests. The linear stiffness and damping were isolated by testing a dry mount and the

effective pumping area was isolated in the testing fixture and plotted as a function of mount

displacement.

Finally, the best method to extract the upper chamber volumetric compliance

parameters was found using a mount with a long narrow inertia track and no decoupler. The

lower chamber volumetric compliance characteristics were only considered to be a function of

preload conditions and were measured using the test apparatus.

The later portion of this chapter focused on the inertia track, decoupler and bell system.

Isolating these components in the experimental fixture enabled direct measurement of flow and

pressure differential, which were then used in the system identification procedures to extract the

parameters. For the inertia track, the inertia and resistance parameters were best extracted using

the least-squares method with random flow perturbations. The decoupler involved a two phase

identification procedure. First, a frequency sweep test was conducted to establish trends in the

uncoupled inertia and resistance. Large amplitude random excitations were then used in the

5.6 Summary

135

Extended Kalman Filter to extract the switching parameter and leak flow resistance. Finally,

the liquid column inertia and resistance associated with the bell were extracted using frequency

sweep measurements. In each of these cases the parameters identified pertained to the CooperStandard mount illustrated in Figure 5.22 and are summarized in Table 5.2.

Table 5.2: Inertia track, decoupler and bell parameters identified for the Cooper-Standard

mount.

Ii

Ri1

Ri2

Id

Rd1

Rd2

=

=

=

=

=

=

6.52 105 kg/s-mm4

4.49 1010 kg/mm7

8.28 108 kg/mm4

4.58 106 kg/s-mm4

4.25 1011 kg/mm7

RL2

Rd3

Ib

Rb1

Rb2

=

=

=

=

=

14.5 mm1

2.27 108 kg/mm4

2.33 108 kg/s-mm4

2.97 1012 kg/mm7

Although the modeling approach throughout this thesis has been to capture the mount

characteristics in as much detail as possible, it may be feasible to reduce model complexity in

some areas.

with the nonlinear resistance model, the decoupler results do not present such a strong case.

The relative magnitude and impact of the free decoupler resistance parameters suggest that the

resistance could perhaps be simplified to a linear form without significant degradation in the

overall system response. However, this is not investigated further as this thesis remains focused

on establishing a complete hydraulic mount model.

Chapter

In this chapter, the nonlinear models developed in Chapter 2 are simulated using the parameters

identified in Chapter 5. Simulation procedures are explained for the inertia track and decoupler

model, as well as the additional bell system.

measured data using dynamic stiffness and phase responses. The models are compared over

the complete frequency and amplitude ranges of interest to validate the modeling and parameter

identification techniques.

6.1

Simulation Techniques

The simulation techniques used to obtain frequency domain data from the nonlinear inertia track

and decoupler model, are presented first. Dynamic system equations, summarized in Section 2.7,

are rewritten in state-space equation form. The continuity equations are

1

Ap X Qi Qd QL

+ R1 Ap X Qi Qd

C1

1

= (Qi + Qd + QL )

C2

P1 =

(6.1)

P2

(6.2)

1

Q i = (P1 P2 (Ri1 + Ri2 |Qi |) Qi )

Ii

1

Q d = P1 P2 Rd1 + Rd2 |Qd | + 1 eRd3 Xd arctan(Qd 2 ) Qd

Id

136

(6.3)

(6.4)

137

V d = Qd

(6.5)

Xd =

Vd

Ap

s

2 |(P1 P2 )|

QL =

arctan ((P1 P2 ) 8 )

RL2_f nc

(6.6)

(6.7)

2

2

Vd arctan( 5 Qd ) Vd_ max 6

RL2_f nc = RL2 1 + 7 7 arctan

Equations (6.1) to (6.7) represent a system of first order differential equations which can

be solved using numerical techniques.

In

the nonlinear model development and parameter identification methods, all variable integration

assumed initial conditions of zero.

jwdr t

conditions be set equal to zero. Since the direct application of input excitation X = Xe

a simulated actuator is required to generate

will give non-zero initial values of both X and X,

the desired input with the appropriate initial conditions.

The single DOF actuator modeled in Figure 6.1 consists of a mass, held in place with a

spring and dash-pot. As illustrated, a force input F (t) on the mass will produce the desired

displacement output X (t) through the equation of motion

+ cX + kX = F

mX

(6.8)

where m, c and k represent the mass, damping and stiffness parameters, respectively. To ensure

the actuator output X(t) is not influenced by the mount being excited, the value of the stiffness

parameter k is several orders of magnitude larger than a typical mount stiffness. This effectively

makes the mount stiffness irrelevant, hence it is not included in equation (6.8). The mass is

138

F(t)

k

m

X(t)

Figure 6.1: Actuator model used to provide excitations with zero initial conditions.

calculated using

m=

k

wn2

(6.9)

where wn is the actuator natural frequency, selected to be 5000 Hz, well above the frequencies

being excited. The damping parameter is selected to achieve critical damping,

c = 2wn m

(6.10)

although it is perhaps not necessary, since the natural frequency is so far beyond the frequency

range of interest. The forced input amplitude F is calculated using the steady-state frequency

response

X

dr

(6.11)

To achieve the desired amplitude of excitation X,

2

mwdr

F (t) = X

+ jcwdr + k ejwdr t

(6.12)

X = X

= F kX cX 1

X

m

(6.13)

(6.14)

139

which is added to the existing differential equations to form a series of seven state equations.

Since the equations (6.1) to (6.5) and (6.13) to (6.14) are in the form y = f (t, y), a numerical

solution can be obtained over a time interval [to , tf ] given initial values yo (to ). For each frequency and amplitude of excitation, the state equations are solved for 10 cycles and sampled at

200 points per cycle. Since the decoupler model introduces small time constants, as illustrated

in Figure 3.10, the equations are considered to be Stiff.

equations are solved by applying a new numerical differentiation technique for stiff differential

equation sets based on Bogacki-Shampine [14]. The numerical differentiation returns the time

domain solution of each state variable in the system. Once the differential equations are solved

, the transmitted force equation is calculated using the time domain state variable solutions, and

equation (2.53)

FT = kr X + br X + (Ap Ad_f nc )(P1 P2 )

+Ap P2 + (Ad Ad_holes ) (Rd1 + Rd2 |Qd |) Qd

where equation (2.52) gives

Ad_f nc

1

= Ad

arctan

2

2

Vd arctan ( 3 P ) Vd_ max 4

and bulge damping R1 (wdr , VT , Fp ) parameters must be correlated with the volume amplitude of

fluid entering the chamber VT . Volume flow into the upper chamber is calculated by integrating

internal flows

VT =

Ap X Qd Qi QL dt

(6.15)

The amplitude of fluid entering the upper chamber VT is then used to update parameters C1 and

R1 , based on the measurements established in Chapter 5. An iterative solution is required to

converge on the appropriate parameters.

140

Once the volumetric compliance C1 and bulge damping R1 parameters have converged,

the frequency domain results of the excitation X and transmitted force FT are calculated via a

Fourier transform. The cross point dynamic stiffness and phase angle relation are calculated with

the method presented in Section 1.2, and the procedure is repeated for each excitation condition

to determine the complete frequency response curve.

To include the bell model in the above simulation procedure, equations (6.1) and (6.2) are

replaced with new continuity equations, from Section 2.7,

(6.16)

C1 P1 = (Am Ab )X Qb Qd Qi QL

(6.17)

C2 P2 = Qi + Qd + QL

(6.18)

Ib

(6.19)

Time domain solutions are obtained using the same numerical differentiation technique.

Transmitted force is also updated to include the influence of the bell chamber pressure, from

equation (2.82)

FT = kr X + br X + (Am Ad_f nc )(P1 P2 ) + Am P2

+(Ad Ad_holes ) (Rd1 + Rd2 |Qd |) Qd (Am Ap )Pb

Volumetric compliance parameters Cb (wdr , VT , Fp ) and Rcb (wdr , VT , Fp ), associated with

the bell, are altered according to the volume amplitude entering the chamber.

Total volume

VT =

Qb (Am Ap Ab )X dt

(6.20)

141

The volume amplitude VT is calculated and used to modify Cb and Rcb . An iterative solution is

again applied to obtain volumetric compliance parameters from measured values.

The solution method developed above represents a computationally intensive numerical

differential solution, coupled with an iterative technique to obtain appropriate system parameters. As a result, the solution time becomes an issue. All simulations conducted for this thesis

were run using MATLAB under Windows, on an Intel PII 233MHz processor. Solutions operating in the nonlinear coupled region required up to five minutes of processing time per frequency

solution, representing a significant number of instructions. However, the processing requirements can be significantly reduced by compiling the code into a stand-alone executable code,

eliminating the run time compiling imposed by MATLAB.

6.2

Model Validation

The Cooper-Standard hydraulic mount investigated in Chapter 5 is now assembled and tested

using the standard dynamic characterization procedure described in Section 1.2.

Measured

dynamic stiffness and phase characteristics are compared with the simulated nonlinear model to

validate the modeling and parameter identification approach. The models are simulated using

parameters identified in Chapter 5 and summarized in Table 5.2. Five system tests include: the

inertia track only, the decoupler only, the decoupler and inertia track, the inertia track and bell,

and the inertia track, decoupler and bell.

6.2.1

Inertia Track

In the first case, the response of the inertia track is observed by removing the decoupler from the

mount in Figure 6.2. The simulated model and measured data are compared in Figure 6.3. Both

the phase angle and dynamic stiffness response of the model is within 5% of measured data.

Several conclusions about the modeling technique and parameter identification can be made.

142

Decoupler

Inertia

Track

Phase (Degrees)

Figure 6.2: Illustration of the Cooper-Standard mount being compared with simulation data.

Frequency (Hz)

1mm P-P

2mm P-P

Figure 6.3: Low frequency simulation versus measured response, on a mount with only an

inertia track.

143

First, the dynamic stiffness and phase characteristics at frequencies beyond the liquid column

resonance demonstrate amplitude dependent behavior. Even without the decoupler, the dynamic

stiffness beyond resonance is higher for the 1 mm P-P (peak-to-peak) amplitude excitation over

the 2 mm P-P response. This nonlinear effect is attributed to the upper chamber rubber stiffness

kr and volumetric compliance C1 parameters. Since the model response captures this effect, the

approach that identifies nonlinear parameters kr , br , C1 and R1 with a local linear response at

each frequency and excitation amplitude, is concluded to be effective. Agreement between the

model and measured response at the resonant frequency, indicates that the inertia track parameter

identification (see Section 5.3) is effective. Also, the phase angle correlation indicates that the

nonlinear resistance model captures the oscillatory fluid damping appropriately.

6.2.2

Decoupler

The Cooper-Standard mount was also assembled and tested with only the decoupler between

chambers. Model simulations under the same conditions are plotted against the measured data

in Figure 6.4.

For both the 0.1 mm P-P and 0.3 mm P-P amplitude conditions, the model

characteristics beyond the liquid column resonant frequency indicate stiffness and compliance

parameters have also been modeled and identified appropriately in the high frequency range.

Perhaps the most significant aspect is the nonlinear decoupler model performance at high

frequencies. Figure 6.4 shows that the amplitude dependent model is within 10% of the measured data. At 140 Hz, a 10% increase in the model damping is attributed to decoupler cage

contact, which introduces a large resistance that halts flow. As explained in Section 3.3, the

decoupler response is pushed into its nonlinear region around resonance at high frequencies.

However, overall the decoupler switching model is effective under high frequency excitations.

Finally, the trends in the 0.1 mm P-P data indicate that the decoupler liquid column resonance is occurring at the correct frequency and is imposing an appropriate amount of damping

144

Phase (Degrees)

Frequency (Hz)

0.1mm P-P

0.3mm P-P

Figure 6.4: High frequency simulation versus measured response, on a mount with only a

decoupler.

on the system. This shows that the identification process has extracted appropriate inertia and

resistance parameter values for the free decoupler.

6.2.3

The mount shown in Figure 6.2, with both the inertia track and decoupler, is now characterized

over the complete range of excitations. Figure 6.5 shows an approximate 5% correspondence

between simulated and measured response in the low frequency range. Comparing these results

to Figure 6.3, it is noted that now the 2 mm P-P excitations yield a higher dynamic stiffness

than for the 1 mm P-P excitations. This indicates how the decoupler dominants the amplitude

dependent characteristics, which have been appropriately modeled and identified. These results

also indicate that the decoupler switching model and parameter identification techniques are valid

under low frequency large amplitude excitations.

145

Phase (Degrees)

Frequency (Hz)

1mm P-P

2mm P-P

Phase (Degrees)

Figure 6.5: Low frequency simulation versus measured response, on a mount including a decoupler and inertia track.

Frequency (Hz)

0.1mm P-P

0.3mm P-P

Figure 6.6: High frequency simulation versus measured response, on a mount including a

decoupler and inertia track.

146

The high frequency system model response in Figure 6.6 is similar to that of a decoupler

alone, and fits the measured data within 5% accuracy. Therefore, the high frequency model with

the decoupler and inertia track is also shown to be effective.

6.2.4

The bell system is now included in the Cooper-Standard mount, as illustrated in Figure 6.7. To

begin, only the inertia track and bell are included in the experimental mount to demonstrate

the influence of the bell system. As Figure 6.8 illustrates, the low frequency model response

falls within 10% of the measured data. The peak in dynamic stiffness at 20 Hz corresponds to

the first natural frequency of the inertia track and bell system. This indicates that the bell has

negligible influence on low frequency characteristics; however, the high frequency response is

significantly altered by the bell plate. As Figure 6.9 indicates, the model and measured response

both capture the notch at 200 Hz. Since this system only includes a bell and inertia track, the

notch corresponds to an antiresonance of the inertia track fluid column, similar to that explained

for the decoupler in Section 3.5. In the high frequency range, the model and measured response

are within 5% agreement. These results indicate that the bell model and parameter identification

procedures have accurately captured the system dynamics.

147

Bell Plate

Bell Position

With Preload

Phase (Degrees)

Figure 6.7: Illustration of the Cooper-Standard mount used to validate system models with a

bell.

Frequency (Hz)

1mm P-P

2mm P-P

Figure 6.8: Low frequency simulation versus measured response, on a mount including an

inertia track and bell system.

148

Phase (Degrees)

Frequency (Hz)

0.1mm P-P

0.3mm P-P

Figure 6.9: High frequency simulation versus measured response, on a mount including an

inertia track and bell system.

6.2.5

149

Finally, the model that includes a bell, decoupler and inertia track is validated. At low frequencies the simulation and experimental data are within approximately 10% deviation, for both the

1 mm P-P and 2 mm P-P excitations. As Figure 6.10 indicates the response is similar to that of

the decoupler and inertia track alone (see Figure 6.5).

The high frequency dynamic characteristics demonstrate the effects of the bell system.

Figure 6.11 shows the amplitude dependent characteristics imposed by the decoupler, as well as

the notch influence of the bell system. However, up to approximately 25% difference between

the model response and measured data is shown. These discrepancies are attributed to a degradation in the upper compliance, occurring in the time between when the parameters were identified

and the bell tests were conducted. The rubber degradation resulted in increased material stiffness, which is demonstrated by the increase in dynamic stiffness at 10 Hz, as compared to those

results shown in Figure 6.6 which were measured several months prior. The decreased compliance has shifted the measured antiresonance frequency to approximately 210 Hz. However, it

is important to focus on how the high frequency trends validate the bell model and parameter

identification techniques.

Figure 6.12 highlights how the bell plate alleviates the high frequency stiffness increase

due to decoupler resonance. Thus, the bell system improves the high frequency characteristics

of a hydraulic mount.

150

Phase (Degrees)

Frequency (Hz)

1mm P-P

2mm P-P

Phase (Degrees)

Figure 6.10: Low frequency simulation versus measured response, on a mount including an

inertia track, decoupler and bell system.

Frequency (Hz)

0.1mm P-P

0.3mm P-P

Figure 6.11: High frequency simulation versus measured response, on a mount including an

inertia track, decoupler and bell system.

151

Phase (Degrees)

Frequency (Hz)

,

Figure 6.12: Comparison of high frequency response with and without the bell plate, using

both measured and modeled response.

6.3 Summary

6.3

152

Summary

This chapter has presented the results of simulation techniques used to obtain frequency domain

model responses with the experimentally identified parameters.

mount, the models are validated with measured response data. The inertia track and decoupler

models, and corresponding parameters, are shown to represent the measured system characteristics. In particular, the decoupler response at high frequencies corresponds with experimental

data and indicates that the nonlinear decoupler model represents the decoupler dynamics.

Small increases in model damping during high frequency cage contact, are attributed to the

resistive flow stopping method imposed by the nonlinear decoupler model. Other discrepancies

are attributed to experimental errors in the measured parameters and general approximations

made with the lumped parameter assumption. However, simulation results presented here are

within acceptable levels by industry standards.

Finally, two mount configurations have been presented that validate the high frequency effects of the bell model. The high frequency notch occurs in both the measured and model

response, thereby validating the modeling approach and explanations regarding the dynamic aspects in Section 3.5.

Having validated the modeling approach and parameter identification procedures, the thesis

will now begin to investigate how some parameters can be associated with mount geometry.

Chapter

This chapter is focused on providing some insight into the relationship between lumped model

parameters and physical component geometry. Specifically, the work presented here investigates

the inertia track, decoupler and bell parameters, using the experimental procedures described

in Chapter 5. The parameters associated with rubber elements are not considered due to the

complex behavior of the material.

7.1

Inertia Track

To confirm trends regarding the inertia track parameters, three test cases are constructed, as

illustrated in Figure 7.1. The different geometry cases include, (a) an inertia track of average

length and cross section, (b) the same inertia track with screws inserted into the fluid column to

increase resistance, and (c) a long inertia track with a small cross section.

7.1.1

Inertia

The inertia associated with the column of fluid in the inertia track has been established in [5, 15,

18, 30]. To relate the inertia to geometry, the inertia track is considered to have a mass equivalent

153

154

L = 212 mm

2

Ai = 57.2 mm

L = 212 mm

2

Ai = 57.2 mm

L = 360 mm

2

Ai = 28.0 mm

Case (a)

Case (b)

Case (c)

to the volume of fluid residing in the inertia track. The inertia becomes

Ii =

Mi

L

=

2

Ai

Ai

(7.1)

where L is the inertia track length and Ai represents the cross-sectional area. The fluid density

in this case is 1.028 106 kg/mm3 (1028 1012 Ns2 /mm4 ) composed of 50% water and 50%

ethylene glycol, by volume.

Using the measured length and cross-sectional area, the inertia parameters are calculated

and compared to those measured experimentally.

tion (7.1) can be used to obtain a good approximation of the inertia parameter based on the

geometry.

Table 7.1: Calculated versus measured inertia parameters for the inertia track.

case (a)

case (b)

case (c)

L (mm)

212

212

360

Ai (mm2 )

57.2

57.2

28.0

Measured

Ii kg/mm4

3.29 106

3.84 106

1.32 105

Calculated

Ii kg/mm4

3.81 106

3.81 106

1.31 105

7.2 Decoupler

7.1.2

155

Resistance

Resistance parameters associated with the inertia track are not as well established. Since these

parameters are influenced by such factors as surface roughness, it is difficult to establish a relation

to the geometry. The resistance parameters Ri1 and Ri2 are identified using the test apparatus

and the identification procedure outlined in Section 5.3. Measured parameters for each test case

are included in Table 7.2.

Table 7.2: Measured resistance parameters of the inertia track.

case (a)

case (b)

case (c)

Ri1 kg/s-mm4

6.52 105

1.09 104

6.14 104

Ri2 kg/mm7

4.49 1010

5.30 1010

1.57 109

The total resistance function Rtotal = Ri1 + Ri2 |Qi | is plotted against the inertia track

flow variable in Figure 7.2a. This resistance is also transformed into velocity damping using Ai .

Pressure loss due to resistance changes from

(7.2)

(7.3)

to

where bi1 = Ri1 A2i , and bi2 = Ri2 A3i . Total velocity damping is plotted in Figure 7.2b and

demonstrates that a reasonable approximation may be to select average damping values based

on experimental results, then convert damping to resistance using Ai . It is recommended that

several more tests be conducted to establish quantitative bounds on the inertia track resistance

and damping parameters.

156

Damping (kg/s)

Resistance (kg/s/mm4)

7.2 Decoupler

Case (a)

Velocity (mm/s)

Case (b)

Case (a)

Case (c)

(a)

Case (b)

Case (c)

(b)

Figure 7.2: Measured resistance parameters for each test case: (a) flow resistance; and (b)

velocity damping.

7.2

Decoupler

The relationship between decoupler parameters and the physical geometry has not been established in any literature reviewed. This study attempts to provide more insight into this relationship by investigating several configurations and their associated parameters.

The inertia

parameter, uncoupled resistance parameters and switching parameters are all considered.

7.2.1

Inertia

The inertia associated with oscillating flow through an orifice is first studied using the two tests

cases illustrated in Figure 7.3.

structed, one with seven 6.3 mm diameter holes and the other with nineteen. Inertia parameters

for each of the two cases are experimentally identified and documented in Table 7.3.

Table 7.3: Effective inertia and fluid column height for a simple orifice.

case (a)

case (b)

Ad holes mm2

218.2

592.3

hf luid (mm)

17.5

20.3

8.24 108

3.55 108

7.2 Decoupler

Cross Section

157

hfluid

Top View

Case (a)

Case (b)

Figure 7.3: Fluid inertia tests of the decoupler orifice with: (a) seven holes; and (b) nineteen

holes.

An interesting property of the measured inertia is the effective fluid column height, denoted

by hf luid . Using the measured orifice inertia Iorif ice and the total area of the decoupler cage holes

Ad_holes , the theoretical fluid column height is calculated with

Iorif ice =

hf luid

Ad_holes

(7.4)

The data summarized in Table 7.3 highlights a curious trend; in both cases the fluid column

height is almost the same. An explanation of this effect involves a detailed look into the fluid

flow behavior and is beyond the scope of this work.

Once the decoupler plate is included, the flow path within the decoupler is altered and the

resulting decoupler inertia changes. Figure 7.4 illustrates how the decoupler plate interrupts

the fluid columns previously investigated. The fluid must now move horizontally within the

decoupler cage to push the decoupler plate. Using this model, a preliminary relationship between

the decoupler features and the effective inertia is developed to be

Id =

Mplate

+ 2

Ad_holes

Ad_holes

(7.5)

7.2 Decoupler

158

hgap

hfluid

where Mplate represents the decoupler plate mass and hgap is the decoupler cage height. This

model assumes that the previously investigated fluid columns that are associated with the simple

orifice, still apply.

However, with the decoupler plate in place, the flow inside the cage is

altered such that short columns of fluid are removed from the original columns. The decoupler

plate mass is considered to contribute to the fluid column inertia, signifying the later term of

equation (7.5).

Two additional test cases are developed by placing rubber decoupler plates within the systems previously investigated and shown in Figure 7.3. Case (c) uses the decoupler cage shown

in case (a) with a rubber decoupler plate. Case (d) adds a rubber decoupler plate to case (b).

Table 7.4 lists the physical geometry parameters associated with the two decoupler cases and

compares the calculated inertia parameters to those measured.

Table 7.4: Calculated versus measured inertia for the complete decoupler system.

case (c)

case (d)

A

d holes

mm2

218.2

592.3

hfluid

(mm)

17.5

20.3

hgap

(mm)

4.8

4.8

Mplate

(kg)

2.36 103

8.59 103

Measured

Id 4

kg/mm

9.87 108

4.14 108

Calculated

Id 4

kg/mm

1.09 107

5.14 108

These tests have achieved a preliminary relation between decoupler geometry and the inertia parameter. More research is required to fully understand the fluid flow patterns and effective

7.2 Decoupler

159

inertia. It should be noted, however, that this research has demonstrated the ability to isolate and

study particular components from the hydraulic mount system. The experimental methods have

opened new possibilities for investigating the internal fluid dynamics. For example, it may be

feasible to perform CFD (Computational Fluid Dynamics) analysis on a system representing the

isolated tests. This technique may prove advantageous to hydraulic mount development, since it

would provide insight into how oscillating fluid flow across an orifice generates an effective fluid

inertia.

7.2.2

Resistance

The decoupler resistance parameters are influenced by numerous physical aspects of the decoupler. Fortunately, the decoupler resistance parameters do not dramatically influence the dynamic

characteristics (see Section 3.1). This study does not attempt to determine a relationship between decoupler geometry and resistance parameters, however resistance data is plotted to highlight trends. Table 7.5 indicates the experimentally measured parameters for each of the test

cases considered in the inertia studies. Figure 7.5 includes plots of the resistance parameters

versus the decoupler flow. The measured results show how the decoupler plate increases the resistance, as flow is forced to change direction inside the cage. Also, cases (c) and (d) indicate a

moderate decrease in resistance as the number of holes increases.

Table 7.5: Measured decoupler resistance parameters.

case (a)

case (b)

case (c)

case (d)

Rd1 kg/s-mm4

1.29 106

5.57 107

4.10 106

3.13 106

Rd2 kg/mm7

2.35 1011

6.51 1012

5.45 1011

4.27 1011

160

Resistance (kg/s/mm4)

7.2 Decoupler

Flow (mm3/s)

Case (a)

Case (b)

Case (c)

Case (d)

Figure 7.5: Measured decoupler resistance parameters plotted for each of the decoupler cases

investigated.

7.2.3

Switching

The final decoupler properties to be investigated are the switching parameters Rd3 and Vd_ max .

These parameters are used in the nonlinear model to limit fluid volume through the decoupler, and

they are also used in the functions Ad_f nc and RL2_f nc . As stated in Chapter 3, these parameters

are related through the decoupler area Ad . To establish this relationship, experimental data is

collected from several decoupler tests, using the identification procedure described in Section 5.4,

and plotted in Figure 7.6. The trendline equation is

Vd_ max

= 15.5 (Rd3 )0.9739

Ad

(7.6)

To obtain the Rd3 parameter given the decoupler geometry, the maximum volume of fluid Vd_ max

is calculated using

Vd_ max =

d

Ad

2

(7.7)

161

Vd_max / Ad (mm)

Parameter Rd3

Measured

Trendline

Figure 7.6: Measured decoupler displacement versus the decoupler switching parameter.

where d denotes the available displacement of the decoupler plate, calculated with

d = hgap hplate

(7.8)

The parameter hplate represents the thickness of the decoupler plate shown in Figure 7.7.

Table 7.6 includes a comparison of the volume amplitude parameters measured from experimental data and calculated from geometry. Comparing the calculated and measured Vd_ max

parameters for cases (c) and (d), it appears as though the experimental identification has picked

up on a thin layer of fluid that slightly reduces the decoupler travel. If 0.5 mm is removed from

hgap the calculated and measured data show improved correlation.

provided only preliminary results, further studies should be conducted to validate the decoupler

parameters.

Table 7.6: Calculated decoupler switching parameters versus measured.

case (c)

case (d)

Ad 2

mm

945

3216

hgap

(mm)

4.8

4.8

hplate

(mm)

2.6

2.6

Measured

Rd3

18.2

19.0

Measured

Vd max

mm3

867

2707

Calculated

Vd max

mm3

1040

3537

162

hplate

hgap

Ad

Figure 7.7: Illustration of decoupler geometry used to calculate the switching parameters.

7.3

Bell Parameters

The bell parameters were not studied due to time constraints, however a few comments can

be made regarding the inertia parameter shown in Table 7.7.

Ib =

hf luid

Ab

(7.9)

The bell fluid column height is found to be within the range of both decoupler tests.

It is

recommended that more effort be placed into investigating this fluid column, similar to the further

studies recommended for the decoupler inertia.

The second point to note is the relatively low resistance parameters associated with the bell

fluid column. As mentioned in Section 3.5, the bell resistance parameters do not have significant

impact on the dynamic characteristics, thus it is not recommended that extensive effort be placed

on developing analytical equations to relate resistance parameters to the geometry.

Table 7.7: Measured bell parameter, indicating fluid column height.

measured

Ab 2

mm

600

Ib 4

kg/mm

2.24 108

Rb1 4

kg/s-mm

2.412 107

Rb2 7

kg/mm

3.74 1012

hf luid

(mm)

13.1

7.4 Summary

7.4

163

Summary

This chapter has provided some preliminary insight into how the inertia track, decoupler and bell

parameters relate to their respective geometry. First, a confirmation has been made regarding

the relation between inertia track geometry and the inertia parameter. The resistance parameters

pertaining to the inertia track are documented, however no conclusive relationship with geometry has been made. Next, the decoupler was investigated. Experimental results for the inertia of

oscillating flow across an orifice were presented and used to establish an effective fluid column

height. A preliminary model for the complete decoupler was developed. The decoupler switching parameters have also been written in terms of the geometry and compared with measured

results. Finally, the bell parameters are documented, but no formal study was conducted.

This chapter has proposed several areas for further research. It is suggested that the inertia

track flow resistance should be converted to velocity damping to establish reasonable relationships with the geometric features. More isolated tests are required to determine the trends.

Resistance to flow across the free decoupler and bell do not have significant impact upon

the dynamic characteristics, thus it is not recommended that extensive effort be placed on developing analytical equations to relate these resistance parameters to the geometry.

The effective fluid column inertia that is associated with oscillating flow through an orifice

needs to be fully understood. Since this thesis has established an effective experimental procedure for identifying parameters, it is now recommended that tools such as CFD be applied to

appreciate the fluid behavior. These techniques will also help to understand the bell fluid column

inertia.

Chapter

The main goal of this thesis was to develop modeling information that can contribute to reducing the hydraulic mount design period. Development time can be significantly reduced with a

reliable fluid mount model that can be simulated to predict the system response before it is physically assembled. Furthermore, it is important for designers to understand how fluid column

dynamics within the hydraulic mount alter the dynamic stiffness and phase characteristics.

The work presented in this thesis builds on recently published hydraulic mount models,

by developing a nonlinear decoupler model that is effective over the complete frequency range

of interest. Further contributions include a model that captures the effect of adding a bell plate

within the upper chamber.

Dynamic stiffness and phase characteristics of the models are explained in terms of the

internal fluid column dynamics.

nonlinear decoupler model, and the linear model that includes the bell plate system.

This thesis also presented a unique experimental approach to identifying hydraulic mount

model parameters. The experimental configuration and identification procedures are presented

to enable an extension of this work to encompass other mount designs. Validation of the models

was accomplished by simulating the equations using experimentally identified parameters and

comparing the results with measured mount data.

164

8.1

Conclusions

165

In conclusion, the models presented within this thesis will provide an excellent tool for

hydraulic mount design. Also, the explanations will give engineers a better understanding of the

internal fluid column dynamics.

8.1

Conclusions

Several conclusions are made that pertain to certain aspects of this work.

The nonlinear decoupler model presented in this work uses flow across the decoupler

in combination with volume and, as a result, is effective over the complete frequency

spectrum of interest. This approach is more effective at high frequencies, since the

decoupler inertia will force the fluid column motion out-of-phase with the pressure

gradient.

A lumped parameter bell model has been established by associating an inertia with the

annular column of fluid between the bell plate and mount wall. Dynamic characteristics

of the bell and decoupler system have been physically explained using a linear, high

frequency, two DOF model. Modal analysis techniques have associated the high

frequency notch, indicative of a bell system, with an antiresonance of the decoupler

fluid column. This high frequency decrease in dynamic stiffness is a function of the bell

chamber compliance, fluid column inertia and effective area parameters within the mount.

Other enhancements to the model include the addition of bulge damping to the upper

chamber. This is accomplished with a resistance parameter on the fluid accommodated by

the volumetric expansion. The parameter is identified and validated in the system models.

8.1

Conclusions

166

Each of the upper chamber compliance parameters were identified through one of three

tests. The stiffness and damping parameters were isolated by testing a dry mount, and

the effective pumping area was isolated in the testing fixture and plotted as a function

of mount displacement. The upper chamber volumetric compliance parameters were

extracted using a mount with a long narrow inertia track and no decoupler. Lower

chamber volumetric compliance characteristics were extracted using the test apparatus

and are only a function of preload conditions.

The inertia and resistance parameters pertaining to the inertia track were best extracted

using the least-squares method with random flow perturbations. However, the decoupler

identification procedure required two separate testing procedures. First, a frequency

sweep test was conducted to establish the uncoupled inertia and resistance parameters.

Large amplitude random excitations were then used with the Extended Kalman Filter

to extract the switching parameter and leak flow resistance. The liquid column inertia

and resistance associated with the bell were also identified using frequency sweep

measurements.

From the system results it is concluded that the approach of identifying nonlinear

parameters kr , br , C1 and R1 with a local linear response at each frequency and excitation

amplitude, is effective.

Agreement between the modeled and measured response also have determined that the

inertia track model and parameter identification are effective. Damping correlation

indicates that the nonlinear resistance model captures the oscillatory fluid damping

appropriately.

8.2 Recommendations

167

At high frequencies, the decoupler response matches experimental data and indicates that

the nonlinear decoupler model represents the decoupler dynamics. Results show how the

amplitude dependent model enters its nonlinear region during larger motions, typically

occurring around resonance at high frequencies. Further results also indicate that the

decoupler switching model and parameter identification techniques are valid under low

frequency large amplitude excitations.

Two mount configurations have been presented that validate the high frequency effects

of the bell model. The high frequency notch occurs in both the measured and model

response, thereby indicating that the modeling approach and the explanations regarding

the dynamic behavior are accurate.

8.2

Recommendations

Recommendations for future research are made regarding the initial work to establish parameterto-geometry relationships.

It is suggested that the inertia track flow resistance parameters can be converted to velocity

damping parameters for a better association with the geometric features. However, more

isolated tests are required to fully establish these trends. Several tests should be conducted

to establish the relationship between damping parameters and general features such as

surface roughness in the internal channel.

The effective fluid column inertia that is associated with oscillating flow through an orifice

needs to be fully understood. Since this thesis has established an effective experimental

8.2 Recommendations

168

procedure, it is now recommended that tools such as CFD be applied to fully appreciate

the fluid dynamic behavior. A CFD model can be constructed to represent the two

chamber test apparatus and validated with measured data. The computational flow

patterns and pressure regions will help to gain a complete appreciation for the lumped

fluid column inertia.

169

Bibliography

[1]

M. Bernuchon. A new generation of engine mounts. SAE Technical Paper Series 840259,

1984.

[2]

Dynamical Systems. CRC Press Inc., Boca Raton, Florida, 1994.

[3]

mount focusing on response to sinusoidal and composite excitations. Journal of Sound

and Vibration, 184:503528, 1995.

[4]

P. Corcoran and G. Ticks. Hydraulic engine mount characteristics. SAE Technical Paper

Series 840407, 1984.

[5]

W. Flower. Understanding hydraulic mounts for improved vehicle noise, vibration and

ride qualities. SAE Technical Paper Series 850975, 1985.

[6]

S. Gau and J. Cotton. Experimental study and modeling of hydraulic mount and engine

system. SAE Technical Paper Series 951348, 1995.

[7]

[8]

M. Ishihama, S. Satoh, K. Seto, and A. Nagamatsu. Vehicle vibration reduction by transfer function phase control on hydraulic engine mounts. JSME International Journal Series C,, 37(3):536541, 1994.

[9]

J.-N. Juang. Applied System Identification. Prentice-Hall Canada Inc., Toronto, 1994.

[10] G. Kim. Study of Passive and Adaptive Hydraulic Mounts. PhD thesis, Ohio State University, 1992.

[11] G. Kim and R. Singh. Resonance, isolation and shock control characteristics of automotive nonlinear hydraulic engine mounts. Transportation Systems, proceedings of

ASME/WAM DSC, 44:165180, 1992.

[12] G. Kim and R. Singh. Non-linear analysis of automotive hydraulic engine mounts. ASME

Journal of Dynamic System Measurement and Control, 115:482487, 1993.

[13] G. Kim and R. Singh. A study of passive and adaptive hydraulic engine mount systems

with emphasis on non-linear characteristics. Journal of Sound and Vibration, 179:427

453, 1995.

170

[14] Lawrence, F. Shampine, and M. Reichelt. The MATLAB ODE suite. Society for Industrial and Applied Mathematics, 18:122, 1997.

[15] K. Lee and Y. Choi. Performance analysis of hydraulic engine mount by using bond

graph method. SAE Technical Paper Series 951347, 1995.

[16] E. Little and R. Kashani. Adaptive-passive and active hydraulic engine mounts. ASME

Advanced Automotive Technologies, DSC-Vol.56/DE-Vol.86:135140, 1995.

[17] Maia, Silva, He, Lieven, Lin, Skingle, To, and Urgueira. Theoretical and Experimental

Modal Analysis. John Wiley and Sons Inc., Toronto, ON, 1997.

[18] D. Margolis and R. Wilson. Modeling, simulation and physical understanding of hydromounts. Masters thesis, University of California, 1997.

[19] The Math Works, Inc., 24 Prime Park Way, Natick MA. MATLAB 5.2, 1998.

[20] R. Matthew and A. Haddow. On the dynamic response of hydraulic engine mounts. SAE

Technical Paper Series 931321, 1993.

[21] MTS Systems Corporation, Eden Prairie, MN 55344-9764. TestStar Installation Manual,

1996.

[22] MTS Systems Corporation, Eden Prairie, MN 55344-9764. TestStar Dynamic Characterization, 1996.

[23] M. Muller, H. Eckel, M. Leibach, and W. Bors. Reduction of noise and vibration by

approximate engine mount systems and active absorbers. SAE Technical Paper Series

960185, 1996.

[24] M. Muller, O. Weltin, C. Freudenberg, D. Law, M. Roberts, and T. Siebler. Engine

mounts and NVH. Automotive Engineering / July, 1994.

[25] D. Rowell and D. Wormley. System Dynamics: An Introduction. Prentice-Hall, Inc.,

Toronto, 1997.

[26] T. Royston and R. Singh. Periodic response of non-linear engine mounting systems. SAE

Technical Paper Series 951297, 1995.

[27] T. Royston and R. Singh. Optimization of passive and active non-linear vibration mounting systems based on vibratory power transmision. Journal of Sound and Vibration,

194(3):295316, 1996.

171

[28] T. Royston and R. Singh. Study of nonlinear hydraulic engine mounts focusing on decoupler modeling and design. SAE Technical Paper Series 971936, 1997.

[29] K. Seto, K. Sawajari, A. Nafamatsu, M. Ishihama, and K. Doi. Optimum design method

for hydraulic engine mounts. SAE Technical Paper Series 911055, 1991.

[30] R. Singh, G. Kim, and V. Ravindra. Linear analysis of automotive hydo-mechanical

mount with emphasis on decoupler characteristics. Journal of Sound and Vibration,

158(2):219243, 1992.

[31] T. Ushijima and T. Dan. Nonlinear b.b.a. for predicting vibration of vehicle with hydraulic

engine mounts. SAE Technical Paper Series 860550, 1986.

[32] T. Ushijima, K. Takano, and H. Kojima. High performance hydraulic mount for improving vehicle noise and vibration. SAE Technical Paper Series 880073, 1988.

[33] F. M. White. Fluid Mechanics. McGraw-Hill, Inc., Toronto, 1994.

[34] W. Wilson. System identification course notes. 1999.

Appendix

The basic idea behind the bondgraph formulation is to keep track of the flow of power within a

physical system. This is accomplished by separating a physical system into components or subsystems. Components in bondgraphs are categorized as 1-port, 2-port and multi-port elements.

It should be noted that there is a difference between the number of physical connections and the

number of ports of an element. A simple example of this is the spring. A spring is connected

between two masses however it can only exchange energy between the spring net displacement

and its force. Thus it is a 1-port element.

In bondgraph, we use two generalized variables for any dynamic systems. These variables are

flow and effort denoted by e(t) and f(t), respectively. The multiplication of the variables in

each system should represent the power flowing between elements. For this reason, the flow and

effort are called power variables. For example, in mechanical systems velocity and force and in

electrical systems current and voltage are selected as the flow and effort variables.

Along with the power variables, there are two other generalized variables called momentum, p(t),

and displacement, q(t), which are called energy variables. The momentum p(t) and displacement

q(t) are defined as the time integral of an effort and flow as

172

p(t) =

e(t)dt

q(t) =

173

f (t)dt

0

Using the energy variables, the power variables can be written as follows:

e(t) = p(t)

f (t) = q(t)

As mentioned earlier, bondgraph elements may have one, two or multiple ports. 1-port elements

include resistance or R-elements, inertia or I-elements and capacity or C-elements. In addition,

there are ideal effort and flow sources that are considered as 1-port elements.

The R-elements generalize the electrical resistance and have a functional relation between the

effort and flow variables as: e = gR (f ). The I-elements generalize the mechanical inertia in the

bondgraph formulation. I-elements have a functional relation between the flow and momentum

variables as: f = gI (p). The C-elements generalize the mechanical springs or electrical capacitors. In this element, there is a functional relation between the effort and displacement variables

as: e = gC (q). Table A.1 shows the functional relation and symboles used for each element in

bondgraph.

Ideal power sources in bondgraphs are ideal effort and flow sources. An ideal effort source is

an element whose effort is only a function of time and independent of its flow. An ideal flow

source is an element whose flow is only a function of time and independent of its effort. Table

A.1 shows the symbols of ideal sources in bondgraphs. It should be noted that in bondgraphs

is used to indicate energy flow while is reserved to indicate information flow.

Transducers or 2-port elements in the bondgraph formulation can exchange energy through two

ports. Electrical motors and hydraulic pumps are examples of transducers. In an ideal transducer

(see TableA.1), there is no loss and the input/output relation is static and linear. The lossless

assumption implies that e1 f1 = e2 f2 and the static and linear assumptions imply that only two

types of tranducers can exist: 1) Transformers where the effort/flow relationships are: e1 = me2 ,

174

Element

Functional Relation

Symbol

R-Element

e = gR (f)

e

f

I-Element

f = gI (p)

e

f

C-Element

e = gC (q)

e

f

Sf

Se

Transducer

Transformer

Gyrator

e1

e1 f1 = e2 f2

e1 = me2 , f1 =

f1

1

m f2

e1 = rf2 , f1 = 1r e2

e1

f1

e1

f1

e1

0-Junction

e1 = e2 = e3 , f1 + f2 f3 = 0

f1

e1

1-Junction

f1 =

1

f

m 2

f1 = f2 = f3 , e1 + e2 e3 = 0

f1

e2

TPT

f2

e2

TF

f2

e2

GY

f2

e3

f3

e2 f2

1

e3

f3

e2 f2

m and r are the transformer and gyrator ratio. In a transformer, an effort (flow) variable is

proportional to an effort (flow) variable via the transformer ratio, while in a gyrator an effort

(flow) variable is proportional to a flow (effort) variable via the gyrator ratio. Examples of

transformers are rigid and massless levers, electrical transformers, gear trains and hydraulic rams.

An example of a gyrator is the ideal electric direct current motor, where the torque is proportional

to the current and the armature voltage is proportional to the rotational speed.

The final bondgraph elements are the multi-port elements or junctions. Similar to transducers,

energy is conserved in junctions. There are only two types of junctions: 1) the 0-junction and, 2)

the 1-junction. In a 0-junction the efforts are equal and the algebraic sum of the flows is always

zero. In a 1-junction on the other hand, the flows are equal and the algebraic sum of the efforts is

always zero. The 1-junction in electrical systems represents Kirchhoffs voltage law written for

175

a loop. The 1-junction for an inertial element is the direct implementation of Newtons 2nd law.

Table A.1 shows the functional relation and symboles used for 0- and 1-junctions.

The final topic of discussion that needs to be addressed is causality. As discussed so far, all

elements exchange energy through their effort and flow variables. In addition, there is a constitutive equation which is the mathematical description of the element and may be dynamic or

static. Because of the constitutive equation one variable is the input or cause and the other is the

output or effect of the element. Since the variables with which any element is interacting in a

bondgraph model are effort and flow, there are two ways of describing an element behaviour, 1)

effort input and flow output or, 2) flow input and effort output. These two situations are called

the causal forms of an element. The graphical symbol used to show causality is a straight vertical line (causal stroke) drawn at one of the ends of energy bond:

or

. The element

nearest the causal stroke has effort impressed on it as input and produces flow as output. Similarly, the element at the other of the bond has flow imposed on it as input and produces effort as

output.

Table A.2: Causality of Bondgraph Elements

Element

Causality

R

R-Element

I-Element

C-Element

Sf

Se

Transformer

Gyrator

0-Junction

1-Junction

TF

TF

GY

GY

176

The causality of sources depend on whether they are a source of effort or flow. In R-elements,

where an effort is related to a flow with an algebraic relation, there is no preference in causality selection as long as the effort or flow can be expressed as a function of each other uniquely.

However, in C- and I-elements the preferred causality is when a flow and effort are the inputs,

respectively. The preferred causality for a C-element is:

This choice of causality is called an integral causality and the outputs (flow and effort) are obtained using integration and not differentiation. Table A.2 shows preferred causality for different

bondgraph elements.

Using the functional equations of transducers shown in Table A.1, the casaulity can be defined.

There are two possible causalities for transformers and gyrators which are shown in Table A.2.

In the 1-junction the flow in all ports is the same and therefore, more than one flow cannot be

defined as being the input. The flow of the other ports will be the outputs of the junction. In

the 0-junction the effort is the same in all ports and therefore, more than one effort cannot be

defined as the input. The effort of the other ports will be the outputs of the junction. Table A.2

summarizes the causality discussion of bondgraph elements.

To this end, we have shown that the modeling of a dynamic system in the bondgraph method

is based upon; 1) identifying all components, 2) connecting them to each other through energy

bonds and 0- and 1-junctions, and 3) assigning causality to each bond using preferred causality.

The last stage in bondgraph modeling it to extract the state-space equations of the system, which

can be obtained by using the following procedure:

Choosing state variables. Each energy storing element (an I- or C-element) with integral

causality has independent initial conditions which gives rise to one state variable. The

number of state variables of a system is equal to the number of energy storing elements

with integral causality. We select the momentum p and displacement q variables of all

177

independent energy storing elements to produce the state vector. The input vector includes

all sources in the system.

Write state equations. Reading from the causal strokes on the bondgraph, write the

constitutive equations of the independent energy storing elements. From the junction that

each independent energy storing element is connected to, write an expression for p and q.

Substitute extra variables. In the p and q expressions, substitute all non-state variables

using the constitutive equations and sources, considering the causal strokes on the

bondgraph.

Appendix

This appendix contains the mathematical development of the frequency domain equations from

the models presented in Chapter 2. The frequency domain equations are used for linear model

simulations presented in Chapter 3. Specifically, the derivations covered here include:

178

B..1

179

The inertia track model is first considered as a linear, large amplitude, low frequency model. The

internal dynamic system is characterized by equations (2.11) to (2.13), repeated here,

C1 P1 = Ap X Qi

(B.1)

C2 P2 = Qi

(B.2)

P1 P2 = Ii Q i + Ri Qi

(B.3)

and the transmitted force to the mount base, from equation (2.14) is,

FT = kr X + br X + Ap (P1 P2 ) + Ap P2

(B.4)

To manipulate the model into a single frequency domain equation, each system variable is converted using the steady-state sinusoidal solution in exponential form:

P1 = P1 ejwdr t

P2 = P2 ejwdr t

(B.5)

i ejwdr t

Qi = Q

jwdr t

X = Xe

Using these transformations, equations (B.1) and (B.2) become

Ap

1

P1 =

X +j

Qi

C1

wdr C1

1

P2 = j

Qi

wdr C2

(B.6)

(B.7)

Ap

X=

C1

Ri + j Ii wdr

1

1

wdr C1 wdr C2

i

Q

(B.8)

180

i gives

and solving for Q

i =

Q

C2 Ap wdr

X

2

Ri C1 C2 wdr + j (Ii C1 C2 wdr

C2 C1 )

(B.9)

The denominator of equation (B.9) represents the characteristic equation of the single degree-offreedom system and can be shown to posses an eigenvalue of

s

(1/C1 + 1/C2 )

wdr =

Ii

(B.10)

if damping is neglected.

Using (B.9) each of the pressure variables can be written in terms of the input amplitude X

P1

P2

Ap

C2 Ap j

=

+

2

2

C

Ri C1 C2 wdr + j (Ii C12 C2 wdr

C2 C1 C12 )

1

Ap j

X

=

2

Ri C1 C2 wdr + j (Ii C1 C2 wdr

C2 C1 )

(B.11)

(B.12)

To obtain the dynamic stiffness characteristics, equations (B.11) and (B.12) are substituted into

the transmitted force (B.4) and then divided by the input amplitude

K (wdr j) =

A2p

C2 A2p j

Ft

= kr + br wdr j +

+

2

C2 C1 C12 )

X

(B.13)

The magnitude and phase angle of the complex equation (B.13) are used to obtain the dynamic

stiffness and phase angle, as follows

K = |K (wdr j)|

= ]K (wdr j)

B..2

The following section presents the development of frequency domain equations for the bell and

decoupler high frequency linear model using modal analysis. The high frequency linear model

181

that includes the decoupler and bell dynamics, from equations (3.19) to (3.23), are:

Cb P b = Qb (Am Ab Ap )X

(B.14)

C1 P1 = (Am Ab )X Qb Qd

(B.15)

C2 P2 = Qd

(B.16)

P1 Pb = Ib Q b + Rb (Qb + Ab X)

(B.17)

P1 P2 = Id Q d + Rd Qd

(B.18)

(B.19)

Equations (B.14) to (B.16) are then substituted into equations (B.17) and (B.18) to yield the two

coupled dynamic equations

1

1

1

A

A

A

A

m

b

m

p

b

b + Rb Q b +

+

Qd =

+

Ib Q

Qb +

X Rb Ab X

C1 Cb

C1

C1

Cb

1

1

1

Am Ab

Id Qd + Rd Qd +

Qb +

+

Qd =

X

(B.20)

C1

C1 C2

C1

These equations are converted to matrix form

Ib 0

0 Id

b

Q

d

Q

Rb 0

0 Rd

Q b

Q d

"

Am Ab

C1

1

C1

+

1

C1

1

Cb

1

C1

Am Ap Ab

Cb

Am Ab

C1

1

C1

+

#

1

C2

Qb

Qd

Rb Ab

0

(B.21)

b ejwdr t

Qb = Q

d ejwdr t

Qd = Q

jwdr t

X = Xe

(B.22)

Substituting these solutions into equation (B.21) yields

1

2

Ib wdr

+ Rb wdr j + C11 + C1b

C1

1

2

Id wdr

+ Rd wdr j +

C1

1

C1

1

C2

182

b

Q

Qd

Fb

Fd

(B.23)

where,

Fb

Fd

"

Am Ab

C1

Am Ap Ab

wdr j

Cb

Am Ab

C1

2

+ Rb Ab wdr

Inverting the coupled matrix (B.23) to solve for the bell and decoupler flow gives,

2

Id wdr

+ Rd wdr j + C11 + C12

C11

Fb

2

b

C11

Ib wdr

+ Rb wdr j + C11 + C1b

Fd

Q

(B.24)

=

d

Q

characteristic equation

where the Characteristic Equation is the determinant of the 2 2 matrix in equation B.23 and

given by

2

det K M wdr

+ Cwdr j =

Ib

Ib

Id

Id

4

2

Id Ib wdr

+

+

+

+ Rb Rd wdr

C1 C2 C1 Cb

1

1

1

+

+

+

C1 C2 C1 Cb Cb C2

3

+ (Ib Rd Rb Id ) wdr

(B.25)

Rb Rb Rd Rd

+

+

+

+

wdr j

C1 C2 C1

Cb

1

+

Rb 0

Ib 0

M=

; C=

; K = C1 1

0 Id

0 Rd

C1

Solving for the bell and decoupler flow yields

2

Rb Ab wdr

(Am Ab )wdr

1

1

2

I

w

+

+

+

R

w

j

+

d

d

dr

dr

C1

C1

C2

(Am Ab )

b =

Q

Characteristic Equation

1

Cb

1

C1

1

C1

1

C2

(Am Ap Ab )

(Am Ab )Cb

+1 j

1

C1

(B.26)

d =

Q

(Am Ab )wdr

C1

1

Cb

Am Ap Ab

(Am Ab )Cb

2

Ib wdr

+ Rb

Characteristic Equation

Rb Ab

Am Ab

wdr j

(B.27)

183

Finally, the cross point dynamic stiffness response function is determined by writing equations

(B.14) to (B.16) in the frequency domain

1

(Am Ab Ap )

Qb +

X

wdr Cb

Cb

Am Ab

1

1

=

X +j

Qb + j

Qd

C1

wdr C1

wdr C1

1

= j

Qd

wdr C2

Pb = j

(B.28)

P1

(B.29)

P2

(B.30)

The dynamic stiffness is then developed from the transmitted force (B.19) in the frequency domain as

1

Ft

d (Am Ap )Pb (B.31)

= kr + br wdr j +

(Am Ad )(P1 P2 ) + Am P2 + Ad Rd Q

X

X

The magnitude and phase angle of the complex equation (B.31) are used to obtain the dynamic

stiffness and phase angle.

Appendix

The following Appendix Includes:

Bill of Materials

Assembly Drawings

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

Appendix

The Extended Kalman Filter algorithm [9, 34] is presented in the following. The algorithm is

separated into five main operations, starting with the nonlinear system representation,

k+1 = f (X

k , Uk , k) + wk

X

k , Uk , k) + qk

Zk = g(X

k and measurement noise qk on the output Zk .

which includes disturbance noise wk on the state X

1.

From the current value of the states at position k, predict the next state values using the state

equations and the system input Uk

pre = f (X

k , Uk , k)

X

k+1

The state vector notation is expanded to clarify the procedure

pre

X1,k+1

f1 (X1,k , X2,k ...Xn,k , Uk , k)

X pre f2 (X1,k , X2,k ...Xn,k , Uk , k)

2,k+1

pre =

X

=

..

..

k+1

.

.

pre

fn (X1,k , X2,k ...Xn,k , Uk , k)

Xn,k+1

For the first iteration, the initial value of the states must be given.

2.

pre

Ck+1

= Fk Ck FkT + wk

203

204

where Ck is the current n n diagonal covariance matrix, or the initial covariance matrix

value for the first iteration. Matrix Fk is a Jacobian matrix of the state equations, evaluated

k

at the current value of the states X

Fk =

f1

X1

f2

X1

f1

X2

f2

X2

..

.

...

...

..

.

f1

Xn

f2

Xn

fn

X1

fn

X2

...

fn

Xn

..

.

..

.

covariance on each state variable.

3.

The Kalman Gain K is next calculated using the predicted covariance matrix

pre

pre T 1

K = Ck+1

GTk qk + Gk Ck+1

Gk

k , Uk , k), evaluated using

The matrix Gk represents the Jacobian of the output function g(X

k

the current value of the states X

Gk =

g

X1

g

X2

...

g

Xn

4.

Using the Kalman Gain, the predicted state values are updated to optimized estimates

k+1 = X

pre + K Zk g(X

pre , Uk , k)

X

k+1

k+1

The Kalman gain acts on the difference between the measured output Zk and the calculated

pre .

output using the function evaluated at the predicted next state values X

k+1

5.

Finally, the predicted covariance matrix is updated using the Kalman Gain

pre

pre

Ck = Ck+1

KGk Ck+1

205

Steps 1 through 5 are repeated for each input and output sample, solving for the best estimate of

k . For each iteration the Jacobian matrices Fk and Gk must be recalculated for

the state matrix X

proper local system linearization.

- Proportional Damping from Experimental DataЗагружено:mat2230
- Structural Dynamics by Mario PazЗагружено:Ravi Poonia
- ME351-MDV-Jan14-DampedFreeVibrationTutorial-Tut02.pdfЗагружено:Jim Harris
- M.tech Sem-2 DYN Tutorial-1Загружено:Rutvik Sheth
- Lecture 07Загружено:Mihir Ashar
- macar200302-13Загружено:JagadeeshMadugula
- AM20120100003_83581247Загружено:Mac Rodge
- AWB120_Dynamics_01_Intro.pdfЗагружено:Ross Waring
- .Nonlinear Dynamics of Mechanical Systems With Friction Contacts by HBMЗагружено:Gerald Sequeira
- Lecture 1Загружено:Lluna Enllena
- Hall-2006_ProblemsEncounteredFromUseRayleighDamping.pdfЗагружено:Oscar Ccama
- 1-s2.0-S1877705811014950-mainЗагружено:Vulebg Vukoic
- ThesisЗагружено:essiengideon
- me147_midterm1_f2011_J._C.Wang_nid298_fid424Загружено:Alberto Lopez
- Mechanical Vibration Forced UndampedЗагружено:Shafiq Shapian
- ME22457_Chapter3_021503_MACLЗагружено:Guillermo Aguilar Hernandes
- DampingЗагружено:gogel26
- VibrationЗагружено:Ana
- FEMCI Book - Miles' EquationЗагружено:பிரபாகரன் ஆறுமுகம்
- Tech Products CatalogЗагружено:kaoblekstena
- Literature ReviewЗагружено:Anis Suraiya
- Dynamic_Loadings_on_Pedestrian_Bridges.pdfЗагружено:Roy Jari
- VibrationЗагружено:Krishna Murthy
- MV TEST II-VVCEЗагружено:Manjunatha Babu N.s
- MIT2_003SCF11_pset10SolЗагружено:Mailson1
- A Parametric Study of Pedestrian Vertical Force Models for DynamiЗагружено:fitg10
- 5thЗагружено:chaitanya mano joel
- mod1Загружено:Suresh Kumar
- Section5_1Загружено:sonti11
- A Review of Design of Shock Absorber Test RigЗагружено:rahul.yerrawar

- Clinical PsychologyЗагружено:Muhammad Affan Fareed
- Chi Squared TestЗагружено:santhi
- Learning LogЗагружено:Jul Biol
- Optimization and Process Parameters of CNC End Milling For Aluminum Alloy 6082Загружено:IJIERT-International Journal of Innovations in Engineering Research and Technology
- 3. JULIE GILUANI ABAO Et Al - Phenomenological Study of How Illness is Socially Cosntructed and Trated Among AntiquenosЗагружено:Josejr Taton
- International Rice Research Notes Vol.24 No.2Загружено:ccquintos
- Effects of Bottom Chord Extensions on the Static and Dynamic Performance of Steel Joist Supported FloorsЗагружено:Jose Prado
- Dr. Eduardo a. Morato Jr Lifelong LearningЗагружено:Sally Java Senayo
- The use of gesture techniques in children’s singing.pdfЗагружено:JessicaKohler
- 1-s2.0-S1048984312000665-mainЗагружено:Ysumaryan Doni
- Diseño Ecologico ExperimentalЗагружено:chechocedros
- AFEM.Ch20.pdfЗагружено:patriciocuevasa
- Medscape Pediatrics Infant COLICЗагружено:Lorena Segovia
- English KAPЗагружено:Adina Ovedenie
- QuestionЗагружено:arghga mandal
- melanie kenton final report - block placementfinishedЗагружено:api-267538132
- CNW- Nov13-2007 - Health Canada Approves Health Claim for HMS 90 - ImmunocalЗагружено:api-3714923
- Cv Atiqi k. Batubara - Volunteer Cifp 2018Загружено:Reynaldy Anantyo
- hazards risk mgmt - session 14 - analyze risk.docЗагружено:Ajesh mohan
- Module 3 Methods of PhilosophyЗагружено:Debbie Marie Bais
- Study of Promotional Strategies of Cadbury in IndiaЗагружено:Ashish Dhingra
- 2184246_1751140911_Austdrinkingwaterguidelines.pdfЗагружено:Atul Tripathi
- Austrias History in Space.pdfЗагружено:Michele Perrone Filho
- Left Out_ Perspectives on Social Exclusion and Inclusion Across Income Groups - ProQuestЗагружено:Scribd
- Models of Policy AnalysisЗагружено:IloNyl
- TocccccЗагружено:Anonymous KJBbPI0xe
- Feasibility StudyЗагружено:Leila Gagap
- EC3099 Industrial economics.pdfЗагружено:magnifiqueheart
- CRM Best Practices Framework ForresterЗагружено:amittsaxena
- Liberation or reproductionЗагружено:Psicología Educativa LUZ-COL

## Гораздо больше, чем просто документы.

Откройте для себя все, что может предложить Scribd, включая книги и аудиокниги от крупных издательств.

Отменить можно в любой момент.