University of Wisconsin Milwaukee

UWM Digital Commons

Theses and Dissertations

December 2013

Techniques for Engine Mount Modeling and

Optimization
Fadi Alkhatib
University of Wisconsin-Milwaukee

Follow this and additional works at: https://dc.uwm.edu/etd

Part of the Mechanical Engineering Commons

Recommended Citation
Alkhatib, Fadi, "Techniques for Engine Mount Modeling and Optimization" (2013). Theses and Dissertations. 344.
https://dc.uwm.edu/etd/344

This Dissertation is brought to you for free and open access by UWM Digital Commons. It has been accepted for inclusion in Theses and Dissertations
by an authorized administrator of UWM Digital Commons. For more information, please contact open-access@uwm.edu.
TECHNIQUES FOR ENGINE MOUNT MODELING AND

OPTIMIZATION

By

Fadi Alkhatib

A Dissertation Submitted in

Partial Fulfillment of the

Requirements for the Degree of

Doctor of Philosophy

in

Engineering

at

The University of Wisconsin – Milwaukee

December 2013
 ABSTRACT

TECHNIQUES FOR ENGINE MOUNT MODELING AND

OPTIMIZATION

By

Fadi Alkhatib

The University of Wisconsin-Milwaukee, 2013

Under the Supervision of Professor Anoop Dhingra

This dissertation presents techniques for the design of engine mounting system to

address the issue of vibration isolation. While the techniques presented herein are

general, the application of proposed techniques is demonstrated primarily through

applications in motorcycles. The dynamic loads that are generated due to the shaking

forces within the engine and the road loads that are transmitted to the engine through the

tire patch are discussed. The geometrical shape of the engine mount is also considered in

this work. All models discussed herein deal with solving the optimization problem for the

engine mount system such that the transmitted forces to and from the engine are

minimized in which the mount parameters are used as design variables.

While work has been done in the past in the area of engine mount design, this

dissertation tries to fill in the gap when it comes to designing a comprehensive mounting

system that takes into account modeling of the mount characteristics, the excitation load

present in the system, and a determination of the final geometrical shape of the engine

mount.

ii
 The work presented in this dissertation discusses three major problems. The first

problem addresses comprehensive mount modeling wherein mathematical mount models

range from a simple Voigt model to a complex Voigt model that captures hysteresis and

nonlinear behavior are presented. The issue of mechanical snubbing is also considered in

these models. An optimization problem is formulated to determine the mount parameters

by minimizing the difference between the transmitted loads predicted by the theoretical

model and experimentally measured values.

The second problem addressed in this dissertation deals with mounting system

optimization. The optimization is carried out such that the loads transmitted through the

mount system from/to the frame are minimized. The road loads that are generated due to

the irregularities in the road profile and the shaking loads that are generated due to the

engine imbalance are discussed in detail. The mount parameters are considered as design

variables. Displacement constraints, both static and dynamic are considered to account

for packaging requirements and to prevent mechanical snubbing of the engine mount.

Numerical examples dealing with mount system optimization are presented first for a six

degree of freedom model that deals only with the powertrain assembly. This is followed

by twelve degree of freedom model that builds on the previous model by considering the

swing-arm assembly dynamics in addition to the powertrain assembly.

The third problem presented in this dissertation deals with finding the optimum

geometrical shape of the mount itself. The shape optimization of the mount is done using

a nonlinear finite element model of the mount developed in ANSYS®. An optimization

problem is formulated to minimize the difference between the target stiffness obtained

from the dynamic analysis and stiffness values obtained from the mount geometry. The

iii
mount geometrical parameters such as the mount diameter and the thickness are used as

design variables. Numerical examples are provided quantifying how mount geometrical

parameters vary for different operating engine speeds.

All the models and techniques developed in this work will help designers

comprehensively design a mounting system that achieves an effective vibration isolation

of the powertrain assembly.

iv
 ACKNOWLEDGMENTS

I would like to thank my advisor, Professor Anoop Dhingra, for his continuous

guidance, patience, support and friendship throughout my dissertation work. I appreciate

my dissertation committee members: Dr Ron Perez, Dr Ilya Avdeev, Dr Nidal Abu Zahra

and Dr Rani El-Hajjar for their support, insights and suggestions. I would like to thank Dr

Sudhir Kual for sharing experimental data used in chapter 3 and for discussing the

general aspects of mount design. I would like also to thank all my colleagues in the

Design Optimization group for their continuous support. I would like also to thank the

University of Wisconsin-Milwaukee for providing me financial support that enabled me

to do this research.

I also would like to thank my father and mother for their influence and support

throughout these long years.

Finally, special thanks to my wife Arwa Shalhout, without her support and

patience this success would not have been possible.

v
Dedicated to my wife Arwa and my Parents who always encouraged me and supported

me throughout these long years of education.

vi
 TABLE OF CONTENTS

ACKNOWLEDGMENTS ................................................................................................ v

Chapter 1 - Introduction .................................................................................................. 1

1.1 Introduction .......................................................................................................... 1

1.2 Vibration Isolation................................................................................................ 2
1.3 Engine Mounts ..................................................................................................... 3
1.3.1 Elastomeric Mounts ...................................................................................... 3
1.3.2 Passive Hydraulic Mounts ............................................................................ 4
1.3.3 Active Engine Mounts .................................................................................. 5
1.4 Modeling of Engine Mounts ................................................................................ 7
1.5 Dissertation Objectives ........................................................................................ 9
1.6 Dissertation Organization................................................................................... 10

Chapter 2 – Literature Review ...................................................................................... 12

2.1 Vibration Isolation.............................................................................................. 12

2.2 Vibration Modes Decoupling ............................................................................. 14
2.3 Mount Modeling................................................................................................. 17
2.4 Load Estimation ................................................................................................. 18
2.5 Shape Optimization ............................................................................................ 20
2.6 Summary ............................................................................................................ 21

Chapter 3 – Mount Characterization and Determination of Mount Parameters ..... 23

3.1 Load Estimation ................................................................................................. 23

3.2 Elastomeric Mount ............................................................................................. 25
3.3 Mount Modeling................................................................................................. 27
3.3.1 Model 1 - Voigt Model ............................................................................... 28
3.3.2 Model 2 - Maxwell-Voigt Model ................................................................ 30
3.3.3 Model 3 - Voigt Model with Bouc-Wen Element ...................................... 31
3.3.4 Model 4 - Voigt Model with Bouc-Wen Element and Nonlinear Stiffness 33
3.4 Parameter Identification ..................................................................................... 35
3.4.1 Numerical example ..................................................................................... 37
3.5 Summary ............................................................................................................ 43
vii
Chapter 4 – Mount Modeling and Design..................................................................... 44

4.1 Mathematical Modeling ..................................................................................... 44

4.1.1 Six DOF Model ........................................................................................... 44
4.1.2 Twelve DOF Model .................................................................................... 51
4.2 Mount Optimization ........................................................................................... 56
4.2.1 Sequential Quadratic Programming ............................................................ 57
4.2.2 Six DOF Model ........................................................................................... 60
4.2.3 Twelve DOF Model .................................................................................... 66
4.3 Isolator Inclining ................................................................................................ 75
4.3.1 Numerical Example .................................................................................... 77
4.4 Vibration Modes Decoupling ............................................................................. 79
4.4.1 Numerical Example .................................................................................... 79
4.5 Summary ............................................................................................................ 85

Chapter 5 – Optimum Design of a Mount System for a V-Twin Engine ................... 87

5.1 Shaking Loads .................................................................................................... 87

5.1.1 Transmitted Shaking Loads ........................................................................ 88
5.1.2 Numerical Example .................................................................................... 98
5.2 Road Loads ....................................................................................................... 103
5.2.1 Transmitted Road Loads ........................................................................... 104
5.2.2 Numerical Example .................................................................................. 106
5.3 Summary .......................................................................................................... 113

Chapter 6 – Shape Optimization of Engine Mounts .................................................. 114

6.1 Finite Element Modeling.................................................................................. 114

6.2 Parameter Optimization.................................................................................... 118
6.3 Design Model and Analysis ............................................................................. 119
6.4 Examples .......................................................................................................... 121
6.4.1 Example I .................................................................................................. 122
6.4.2 Example II................................................................................................. 126
6.5 Summary .......................................................................................................... 129

Chapter 7 - Conclusion ................................................................................................. 130

viii
 7.1 Theoretical Modeling ....................................................................................... 130
7.2 Mounting system optimization ......................................................................... 131
7.3 Shape Optimization .......................................................................................... 132
7.4 Future Work ..................................................................................................... 133

References ...................................................................................................................... 135

APPENDIX A ................................................................................................................ 140

APPENDIX B ................................................................................................................ 142

APPENDIX C ................................................................................................................ 144

APPENDIX D ................................................................................................................ 146

ix
 LIST OF FIGURES
Figure 1.1: Mechanical Model for Elastomeric Mounts ..................................................... 4

Figure 1.2: Simple Hydraulic Mount .................................................................................. 5

Figure 1.3: (a) Mechanical Model for Active Elastomeric Mount, (b) Mechanical Model

for Active Hydraulic Mount................................................................................................ 6

Figure 1.4: Engine Six DOF Modes (Ye, et. al. 2001) ....................................................... 8

Figure 1.5: (a) Engine Mount, (b) Tri-Axial Engine Mount Model (Kaul, 2006) .............. 8

Figure 3.1: Schematic Diagram Showing the Tire Patch .................................................. 24

Figure 3.2: Tri-axial Engine Mount Model....................................................................... 27

Figure 3.3: Voigt Model ................................................................................................... 29

Figure 3.4: Maxwell-Voigt Model .................................................................................... 31

Figure 3.5: Voigt Model with Bouc-Wen Element ........................................................... 33

Figure 3.6: Voigt Model with Bouc-Wen Element and Nonlinear Stiffness .................... 34

Figure 3.7: Measured Force – Displacement Curve. ........................................................ 38

Figure 3.8: Force Displacement Curve (Voigt Model) ..................................................... 38

Figure 3.9: Force Displacement Curve (Maxwell-Voigt Model) ..................................... 39

Figure 3.10: Force Displacement Curve (Voigt Model with Bouc-Wen Element) .......... 39

Figure 3.11: Force Displacement Curve (Voigt Model with Bouc-Wen Element and

Nonlinear Stiffness) .......................................................................................................... 40

Figure 4.1: Six DOF Model (Cocco, 2001) ...................................................................... 45

Figure 4.2: Twelve DOF Model (Cocco, 2001) ................................................................ 54

Figure 4.3: Twelve DOF Engine-Swingarm Layout......................................................... 55

Figure 4.4: Mount System Layout .................................................................................... 63

x
Figure 4.5: Mode Shapes for the Optimized Configuration (6 DOF Model) ................... 64

Figure 4.6: Mode Shapes - 1 to 6 (12 DOF Model – Shaking Load only) ....................... 69

Figure 4.7: Mode Shapes - 7 to 12 (12 DOF Model – Shaking Load only) ..................... 70

Figure 4.8: Road Profile.................................................................................................... 72

Figure 4.9: Mode Shapes - 1 to 6 (12 DOF Model – Combined Loading) ....................... 73

Figure 4.10: Mode Shapes - 7 to 12 (12 DOF Model – Combined Loading) ................... 73

Figure 4.11: Schematic diagram of equipment supported by inclined isolators ............... 76

Figure 4.12: Mount Arrangement ..................................................................................... 77

Figure 4.13: Decoupled Modes Along with the Associated Natural Frequency .............. 78

Figure 4.14: Decoupled Modes and the Corresponding Frequencies (Case I) ................. 84

Figure 4.15: Decoupled Modes and the Corresponding Frequencies (Case II) ................ 84

Figure 4.16: Decoupled Modes and the Corresponding Frequencies (Case III)............... 85

Figure 5.1: Slider Crank Mechanism ................................................................................ 89

Figure 5.2: V-twin Engine Configuration. ........................................................................ 92

Figure 5.3: Free Body Diagram of the Piston ................................................................... 96

Figure 5.4: Shaking Force in the x and y Directions (800 rpm) ....................................... 99

Figure 5.5: Shaking Force in the x and y Directions (3000 rpm) ..................................... 99

Figure 5.6: Shaking Force in the x and y Directions (5000 rpm) ................................... 100

Figure 5.7: Shaking Torque (800 rpm) ........................................................................... 100

Figure 5.8: Shaking Torque (3000 rpm) ......................................................................... 101

Figure 5.9: Shaking Torque (5000 rpm) ......................................................................... 101

Figure 5.10: Road Profile #1 ........................................................................................... 107

Figure 5.11: Magnitude Plot of the Spectrum for Road Profile #1 ................................. 107

xi
Figure 5.12: Reconstructed Time Plot for Road Profile #1 ............................................ 108

Figure 5.13: Road Profile #2 ........................................................................................... 109

Figure 5.14: Magnitude Plot of the Spectrum of Road Profile #2 .................................. 109

Figure 5.15: Reconstructed Time Plot for Road Profile #2 ............................................ 110

Figure 6.1: Mount Geometry .......................................................................................... 115

Figure 6.2: Solid 186 Element (ANSYS, 2009) ............................................................. 116

Figure 6.3: Isometric View Showing the Boundary Conditions ..................................... 116

Figure 6.4: Front View Showing the Boundary Conditions ........................................... 117

Figure 6.5: Isometric View Showing the Constraint Equation ....................................... 117

Figure 6.6: Front View Showing the Constraint Equation ............................................. 118

Figure 6.7: Isometric View of the Initial Geometry ....................................................... 123

Figure 6.8: Front View of the Initial Geometry .............................................................. 123

Figure 6.9: Isometric View of the Optimized Geometry (Example I) ............................ 124

Figure 6.10: Front View of the Optimized Geometry (Example I) ................................ 124

Figure 6.11: Design Variables Vectos Vs. Steady State Speed ...................................... 125

Figure 6.12: Isometric View of the Optimized Geometry (Example II)......................... 127

Figure 6.13: Front View of the Optimized Geometry (Example II) ............................... 128

xii
 LIST OF TABLES
Table 3.1: Computed Parameters for the Four Models ..................................................... 41

Table 4.1: Bounds for Design Variables ........................................................................... 64

Table 4.2: Inertia Tensor of Powertrain Assembly ........................................................... 64

Table 4.3: Optimization Results (6 DOF Model) ............................................................. 65

Table 4.4: Optimization Results for Position and Orientation (6 DOF Model)................ 65

Table 4.5: Inertia Tensor of the Swing-arm Assembly ..................................................... 68

Table 4.6: Optimization Results (12 DOF Model – Shaking Load only) ......................... 68

Table 4.7: Optimization Results for Position and Orientation (12 DOF Model – Shaking

Load only) ......................................................................................................................... 69

Table 4.8: Optimization Results (12 DOF Model – Combined Loading) ........................ 71

Table 4.9: Optimization Results for location and Orientation (12 DOF Model –

Combined Loading) .......................................................................................................... 72

Table 4.10: Stiffness Values and the Distance.................................................................. 77

Table 4.11: Inertia Tensor for the Engine ......................................................................... 80

Table 4.12: Initial Guess of the Mount Locations in (m) ................................................. 80

Table 4.13: Initial Guess of the Mount Orientations (deg.) .............................................. 81

Table 4.14: Initial Guess of the Mount Stiffness (N/m) ................................................... 81

Table 4.15: Optimum Mount Stiffness (Case I)................................................................ 81

Table 4.16: Stiffness Matrix Showing Off-Diagonal Terms (Case I) ............................... 82

Table 4.17: Optimum Mount Stiffness (Case II) .............................................................. 82

Table 4.18: Stiffness Matrix Showing Off-Diagonal Terms (Case II) ............................. 82

Table 4.19: Optimum Mount Stiffness (Case III) ............................................................. 83

xiii
Table 4.20: Stiffness Matrix Showing Off-Diagonal Terms (Case III) ............................ 83

Table 5.1: Shaking Force Vector at Different Speeds .................................................... 102

Table 5.2: Optimization Results for the Shaking Force Vector Corresponding to 800 rpm

......................................................................................................................................... 102

Table 5.3: Optimization Results for the Shaking Force Vector Corresponding to 3000

rpm .................................................................................................................................. 102

Table 5.4: Optimization Results for the Shaking Force Vector Corresponding to 5000

rpm .................................................................................................................................. 103

Table 5.5: Optimization Results for Road Profile #1 ..................................................... 110

Table 5.6: Optimization Results for location and Orientation (Profile #1) .................... 111

Table 5.7: : Optimization Results for Road Profile #2 ................................................... 111

Table 5.8: Optimization Results for location and Orientation (Profile #2) .................... 112

Table 6.1: Parameter Optimization Results .................................................................... 125

Table 6.2: Design Variables Vector Corresponding to different Steady State Speeds ... 126

Table 6.3: Transmitted Loads at Different Operating Speeds ........................................ 126

Table 6.4: Parameter Optimization Results .................................................................... 128

xiv
 1

1 Chapter 1 - Introduction

The mounting system is the primary interface between the powertrain and the

frame; therefore, it’s vital to the determination of the vibration isolation characteristics.

Different types of engine mount are presented in this chapter, but the only engine mount

that will be used in the work herein are the elastomeric mounts. The elastomeric mounts

are made of rubber which withstands large amount of deformation under loads with the

ability to almost retain its original shape when the load is removed. This is due to the

inherent material property of rubber. Rubber is a viscoelastic material which enables it to

be used as an isolator and as a damper.

1.1 Introduction

There are two major problems that engineers must deal with when it comes to

vibration isolation. The first problem is force isolation, which is frequently encountered

in rotating or reciprocating machinery with unbalanced masses. The main objective in

this problem is to minimize the force transmitted from the machine to the supporting

foundation. The second problem is motion isolation. In this case, we are interested in

minimizing the transmitted vibration amplitude such that the mounted equipment is

shielded from vibrations coming from the supporting structure. This is broadly achieved

by mounting equipment on a resilient support or an isolator such that the natural

frequency of the equipment-support system is lower than the frequency of the incoming

vibrations to be isolated.
 2

1.2 Vibration Isolation

Vibration isolation can be simply defined as isolating an object from the source of

vibration. In order for this objective to be achieved, isolators must be used. There are two

major types of isolation. The first type is the passive isolation in which passive

techniques such as rubber pads or mechanical springs are used. The isolation is achieved

by limiting the ability of vibrations to be coupled to the structure being isolated. This is

done using a mechanical connection which dissipates or redirects the energy of vibration

before it gets to the structure to be isolated. Passive methods sometimes involve

electromechanical controls for adjusting the system, but the isolation mechanism is

passive. Passive systems are cost effective and their relative simplicity makes them more

reliable and safe. Elastomers which are used in the automotive industry to isolate the

engines are one of the most widely used passive isolators.

The second type of isolation is active isolation which contains along with the

spring, a feedback circuit which consists of a sensor such as a piezoelectric

accelerometer, a controller and an electromagnetic transducer. The acceleration

(vibration) signal is processed by a control circuit in which is feed to the electromagnetic

actuator which amplifies the signal. As a result of such a feedback system, a considerably

stronger suppression of vibration is achieved compared to the ordinary damping. Most

active vibration isolation systems are relatively complex, costly, and often provide only

marginal improvements in performance compared with conventional passive vibration

isolation techniques. They are also more difficult to set up, and their support electronics

often require adjustment. Nonetheless, active systems can provide function which is

simply not possible with purely passive systems. However, due to their cost
 3

effectiveness, reliability and relative simplicity passive isolators will be used in the

design of the mounting system throughout this study.

1.3 Engine Mounts

To achieve the best vibration isolation for the powertrain, a mounting system is

used to mount the powertrain in place. The mounting system will provide isolation that

will in turn minimize the transmitted forces to/from the engine to the frame. On the other

hand, it will also prevent engine bounce caused from shock excitation. This goal is

achieved by making the dynamic stiffness and damping of the mounting system

frequency and amplitude dependent. Three different types of engine mount systems are

listed below:

1.3.1 Elastomeric Mounts

Elastomeric mounts, which are made of rubber, have been used to isolate engines

since 1930s. A lot of changes have been made over the years to improve the performance

of the elastomeric mounts. For proper vibration isolation, elastomeric mounts are

designed for the necessary elastic stiffness rate characteristics in all directions. They are

maintenance free, cost effective and compact. The elastomeric mounts can be represented

by a Voigt model which consists of a spring and a viscous damping as shown in Fig. 1.1.

It is difficult to design a mounting system that satisfies a broad array of design

requirements. A mount with high stiffness or high damping rates can yield low vibration

transmission at low frequency, but its performance at high frequency might be poor. On

the other hand, low stiffness and low damping will yield low noise levels but it will

induce high vibration transmission. A compromise is needed to obtain balance between

 4

engine isolation and engine bounce. In order to achieve low vibration transmissibility, the

mount stiffness must be as low as possible. However, this causes increased static

deflection. Lower damping is also desirable for lower transmissibility at higher frequency

range. One the other hand, handling and maneuverability are enhanced with higher

stiffness. Elastomeric mounts provide a trade-off between competing requirements of low

static deflection and enhanced vibration isolation.

x f

k c

Figure 1.1: Mechanical Model for Elastomeric Mounts

1.3.2 Passive Hydraulic Mounts

Hydraulic mounts were first introduced in 1962 for use as vehicle mounting

systems. Since then, their popularity has improved for two reasons. The first one is that

the current vehicles tend to be small, lightweight and front wheel drive with low idle

speeds. The second one is that the hydraulic mounts have developed into highly tunable

devices. Three types of hydraulic mounts are in use these days and these are: hydraulic

mount with simple orifice, hydraulic mount with inertia track, and hydraulic mount with

inertia track and decoupler. A general schematic diagram of the hydraulic mount is
 5

shown in Fig. 1.2. Although there are differences between orifice and inertia track

mounts, all of them cause damping at low frequency ranges. These mounts can be tuned

to have high damping at the shock excitation frequency which is used to reduce the

vibration levels. The dynamic stiffness of these mounts is usually higher than that of the

elastomeric mounts. Although the damping in these mount is high at low frequency, the

isolation at higher frequencies is degraded. This problem is handled by adding a

decoupler to the hydraulic mount which operates as amplitude limited floating piston. It

allows the mount to behave like an elastomeric mount to provide good vibration isolation

at large displacement. On the other hand, it allows it to behave like a normal hydraulic

mount providing the damping for shock excitation.

Rubber

Liquid Orifice

Secondary Rubber
Air

Figure 1.2: Simple Hydraulic Mount

1.3.3 Active Engine Mounts

In active vibration control, a counteracting dynamic force is created by one or

more actuators in order to suppress the transmission of the system disturbance force. A

general active mount consists of a passive mount (elastomeric or hydraulic), force

 6

generating actuator, a structural vibration sensor and an electronic controller. The passive

mount is used to support the structure in case of an actuator failure. The controller can

either be feedback or feed forward. The vibration control is implemented with a closed

loop controller that utilizes the sensor measurement. The mechanical models of

elastomeric and hydraulic active mounts are shown in Fig. 1.3. The active mount stiffness

is equivalent to the stiffness of the passive mount (elastomeric or hydraulic). The active

mounts can overcome the limitations of passive mounts. Active elastomeric mounts can

be very stiff at low frequencies and very soft at high frequencies. Meanwhile the active

hydraulic mounts can be tuned to achieve adequate damping at engine bounce frequency

and have very low dynamic stiffness at high frequency. Semi active mounts are used to

improve the low frequency features of the system like increasing damping. By providing

superior isolation, active engine mounts can allow large engine vibration levels. This may

reduce balance shaft requirements and enable the vehicle chassis to be lighter.

Structure Structure

F
𝐾𝐾2

K M
C F K C

Engine Engine
(a) (b)

Figure 1.3: (a) Mechanical Model for Active Elastomeric Mount, (b) Mechanical Model
for Active Hydraulic Mount.
 7

1.4 Modeling of Engine Mounts

A typical mounting system consists of powertrain and a number of mounts that

connect the powertrain to the supporting frame. The major objective of the engine mount

is to isolate the engine disturbances from being transferred to the supporting structure.

These disturbances will excite the engine 6 DOF vibration modes shown in Fig. 1.4. For

example, the torque caused by the firing pulse will cause engine pitch vibration. To

isolate vibrations caused by engine unbalanced disturbances, low elastic stiffness as well

as low damping are used since the transmitted forces depends on the values of the

stiffness and damping of the mounts. The mounts are modeled as a spring and hysteresis

damping or viscous damping along each of the three principal directions shown in Fig.

1.5. The mounts used herein are elastomeric mounts in which the stiffness, orientation

and location are the main variables that need to be determined in order to achieve the

desired isolation. This type of engine mount is modeled as Voigt Model which is shown

in Fig. 1.1. The frame is always modeled as a rigid body thorough out this dissertation.

The natural frequency of the mounting system should be lower than the engine

disturbance frequency to avoid the excitation of the mounting system resonance. This

will ensure a low transmissibility.

 8

Figure 1.4: Engine Six DOF Modes (Ye, et. al. 2001)

Figure 1.4, shows an engine and its six degrees of freedom that will be excited as

result of the inertia forces acting on the its block and the oscillator torque acting about the

crankshaft. Fig. 1.5, shows a typical engine mount alongside its tri-axial model consisting

of a spring and damper along each principal direction.

y
z
x

(a) (b)

Figure 1.5: (a) Engine Mount, (b) Tri-Axial Engine Mount Model (Kaul, 2006)
 9

1.5 Dissertation Objectives

The emphasis in this work is to develop a complete mounting system such that the

transmitted forces to/from the engine to the supporting frame are minimized. The loads

can be described as external or internal. The external loads are due to external

disturbances from the environment such as irregularities in the road profile and road

bumps. These disturbances can be transmitted through the tire patch to the engine causing

it to hit nearby components. The engine movement needs to be constrained due to

packaging space limitations surrounding the engine. In order to do so, these transmitted

external loads must be minimized by the use of the mounting system. Once the mounting

system is defined and the transmitted loads are minimized, the focus is switched to the

balancing masses inside the engine. This is done to minimize the internal loads. The loads

that need to be minimized are the loads due to the shaking force resulting from the

rotating unbalance due to eccentric masses. Material imperfections, faulty assembly and

machining inaccuracy are among other factors that will cause eccentric masses. This will

introduce an offset between the center of gravity and the axis of rotation leading to

unbalanced forces. These forces, which vary in magnitude and direction, can be

eliminated by introducing a counter force that eliminates the effect of the original

unbalanced force.

In order to minimize the transmitted external loads to the engine, we first need

good load estimates. The technique used herein is to estimate the force transmitted

through the tire patch for different load profiles. The next step after defining the external

transmitted loads is to design the mounting system. Two different performance metrics

will be used to solve this problem. The first one involves decoupling the vibration modes
 10

through minimizing the off-diagonal terms of the global stiffness matrix. This insures that

only the diagonal terms in the stiffness matrix are dominant, which leads to clean

decoupled modes. The second method deals with minimizing the transmitted loads.

Several alternate mount models will be explored. These include the Voigt model,

Maxwell Voigt model, Voigt model with a Bouc-Wen element and Voigt model with

Bouc-Wen element and a variable stiffness. All of these models will have a component

that represents mechanical snubbing. Mechanical snubber is used to absorb large amount

of energy within small displacement amplitudes. In order to capture the hysteretic

behavior over large range of operating frequencies, a Bouc-Wen model is added.

After fully defining the mounting system, as a final step, an optimum geometrical

shape of the mount will be determined using topology optimization.

1.6 Dissertation Organization

This work has been divided into five main parts. Chapter two discusses the

available literature on vibration isolation, vibration modes decoupling, mount design and

shape optimization.

Chapters three and four present the work regarding mount modeling and design.

These chapters provide the necessary mathematical modeling along with the equations of

motion for various mount models used in this work. Numerical examples utilizing

experimental mount data are presented to demonstrate extraction of mount parameters

from solution of an optimization problem.

Chapter five discusses the issue of the load transmitted from/into the engine and

the balancing masses. All the mathematical formulation for the internal forces and

moments are presented along with numerical examples. External loads imposed on the
 11

system due to irregularities in the road profile are discussed and different road profiles

are analyzed. This road profile information is used to determine the force transmitted the

tire patch in the vehicle.

Chapter six discusses mount shape optimization problem providing examples that

ties it to the mount modeling and design discussed in the previous chapters showing the

effect of different engine operating speeds on the final mount shape. A nonlinear finite

element analysis is performed to determine the optimum mount shape.

Chapter seven summarizes the main results and conclusion of the dissertation and

provides an outline for possible future work.

 12

2 Chapter 2 – Literature Review

Over the years, much work has been done in the area of vibration isolation.

Throughout this thesis, the focus is on vibration isolation through the use of a mounting

system. There are various methods that have been used to minimize vibration transmitted

to and from the engine. Among these techniques, mount system optimization stands out.

The mount optimization problem typically involves finding the optimum stiffness,

orientation and location of the mounting system that will result in the best possible

vibration isolation. Once the necessary mount characteristics are known, the problem of

finding the optimum geometrical shape of the engine mount is also considered in this

thesis.

2.1 Vibration Isolation

Spiekermann, et al. (1985) discussed the issue of minimizing forces that are

transmitted through the mounting system. These forces can be caused as a result of

rotational imbalance and reciprocating masses. The authors argue that in the case of small

damping and frequencies below the natural frequency, the force transmitted through the

mounts is proportional to the mount stiffness. Nevertheless, when the excitation

frequency is near the rigid body natural frequency, the rigid body displacement and the

transmitted forces may be large. The procedure used in the optimization technique is

removing the natural frequencies form the undesired range and keeping the others. This is

done by using an objective function in the optimization procedure that penalizes the

natural frequencies in the undesired range without affecting other design parameters. For

simulation purposes, a three dimensional rigid body is used. The rigid body consists of
 13

six degrees of freedom (DOF) that includes three translational DOF and three rotational

DOF.

Ford (1985) presents a design procedure for the front wheel drive engine idle

isolation. In this procedure, a six degree of freedom lumped system is used to represent

engine mounts. Then decoupling the highest five natural frequencies from the idle torque

pulse direction is achieved. The baseline mounting system and the decoupled mounting

system are tested on a three cylinder engine with similar inertia properties to the four

cylinder engine. The main disturbance at the idle is the crankshaft torque vibration caused

as a result of the gas pressure firing pulse. The approach is to decouple the torque

generated by gas pressure pulse from five of six powertrain rigid body modes. This is

done by introducing an objective function which is the sum of the square of the roll

component in modes two through six.

Sui (2003) emphasized on the role of mounts in achieving better vehicle handling

characteristics and rider comfort as well as a resulting vibration caused by engine firing

force and other sources. This is achieved only when there is a mounting system that

exhibits decoupled vibration modes. In order to achieve decoupling, the following

assumptions must be considered. The powertrain is infinitely rigid and mounted to the

ground. The excitations are assumed to be of a harmonic or periodic nature with known

frequencies and the resulting displacements are small. The author lays down some basic

concepts that include the following definitions of different coordinate systems: the

vehicle coordinate system, engine coordinate system, principal moment of inertia (MOI)

coordinate system, torque roll axis coordinate system, elastic axes and elastic center and

center of percussion.
 14

2.2 Vibration Modes Decoupling

Six vibration modes will be generated for the engine mass since it possesses six

degrees of freedom in 3-D space. Three of the modes are translational and three are

rotational. Generally speaking, the shaking force will cause the engine to respond in six

degrees of freedom. The work is done to decouple the modes or make the coupling weak.

This technique will be used as one of the proposed methods to optimize the mounting

system.

Timpner (1965) suggested different techniques to eliminate vibrations due to

internal and external disturbances. In order to decouple the modes, the elastic center of

the mounts must coincide with the center of gravity of the engine. As a result, the ideal

locations of the mounts are inside the mass (engine) which is infeasible. Luckily, the

mounts can be still placed outside the engine and still achieve the goal of having the

elastic center and the center of gravity coincide. The author discusses three different

engine mounts orientations. First: two equal mounts symmetrically located. Second: two

equal mounts with axes normal to each other. Third: two vertical mounts with different

rates

Liu (2003) presents a method used in the optimization design of engine mounts.

The constraint problem is solved using some of the known parameters such as engine

center of gravity, mount stiffness rates and mount location and/or orientation. The main

objective of this work is decoupling vibration modes. This work is done using a computer

code DynaMount. Generally speaking, it’s hard to come up with a mount design that

decouples vibration modes. However, there are few special cases in which vibration

modes can be completely decoupled. Throughout the study, the author used two different
 15

coordinate systems. The first one is the vehicle coordinate system which is located at the

engine center of gravity, and the equation of motion of the system is written with respect

to it. The second one is a local coordinate system used to describe the engine mount

properties. A rotation matrix that relates the local coordinate system to the global

coordinate system is used in this case. The mounts are considered to be cylinders with the

top surface attached to the engine and the bottom surface attached to the vehicle. The

origin of the local coordinate system is located at the center of the mount.

Jeong and Singh (2000) examined the issue of torque roll axis (TRA) decoupling

for a multi-dimensional mounting system of an automotive engine and gear box. They

consider only the rigid body modes of the powertrain and the chassis is considered to be

rigid. Since pulsating torque of the multi cylinder engine is a major source of vibration,

therefore a mathematical model of the engine mounting system necessary to understand

the design issues.

Iwahara and Sakai (1999) discussed various possibilities to isolate the engine. The

engine mount layout consists of four mounts supporting the engine. The three and five

mount layouts among other layouts are also investigated. Eigen value analysis, frequency

response and transient response are used to determine the best way to isolate the engine.

Derby (1973) presents two techniques for decoupling. The first one is locating the

isolators symmetrically in the same plane with the center of gravity. The second one is

locating the isolators symmetrically about a ring in which the center of gravity is higher

than the center of the ring. The author presents the necessary condition to decouple the

translational modes from the rotational modes as well as decoupling natural frequencies.

The isolators are located at the corners of a plane rectangle and the center of the
 16

equipment is located within the rectangle formed by the isolators. In the paper, it’s

assumed that the damping matrix is proportional to the stiffness matrix. Furthermore, it’s

assumed that the stiffness values for all isolators are equal. Finally, the center of gravity

is located above the center of the rectangle pattern of the isolators which reduces the

number of equation to only two instead of eight, and the number of parameters to five.

Akanda and Adulla (2005) studied a six cylinder four wheel drive vehicle. In such

a vehicle, the powertrain includes engine, transmission and transfer case. The torque roll

axis approach is used to decouple the modes and come up with the mounting system

locations. The author suggests locating the mounts at the nodal points of the fundamental

bending modes of the powertrain may reduce the transmitted forces to the body.

Bretl (1993) presents a new simulation method to design the mounting system.

The author sets the goal to come up with a mounting system that minimizes the response

regardless of the resulting rigid modes. The technique computes response sensitivity to

determine changes to the mounting system that will result into a minimum response. The

design variables are the mount location, stiffness and damping. The response sensitivities

are used to construct a set of linear equations that represent the total difference in

response between the target and computed as a summation of design variable changes.

The updated factors are approximated to the design variables that are required to

minimize response.

Courteille and Mortier (2005) present a new technique to find an optimized and

robust solution for the mounting system. Multi objective algorithm (Pareto optimization)

is used as a base to the multi objective robust optimization problem. The use of this

technique enhances the vehicle isolation characteristics. The method focuses on the use
 17

of the optimization to minimize the vibration due to pulsating torque of the crankshaft at

idle speed without paying any attention to the unbalanced forces due to the forces of the

engine pistons. Since there is no accurate definition of the vehicle at early stages of

design process, the author uses a probability distribution of the system parameters. This

leads to a random change of the system’s parameters. In order to have a sound design for

an engine mount that will perform the intended job in isolation, a good estimate of the

loads acting on the structure is very important. Reviewed next are some methods that can

be used to estimate the loads acting on the mount system.

2.3 Mount Modeling

The first step in mount design is modeling of the mount itself. Simple Voigt

model is frequently employed to model the mount. The model consists of a spring and a

damper connected in parallel and supporting the isolated mass. While the Voigt model is

sufficient in many applications, it cannot capture certain mount characteristics such as

hysteresis behavior, mount snubbing when shock loading is present, nonlinearity in

mount systems, etc.

Zhang and Richards (2006) presented a study of the dynamic analysis and

parameter identification of a rubber isolator using Maxwell-Voigt model. In the study,

they noticed the difference between the Voigt model which simply consists of a spring

damper connected in parallel and the Maxwell-Voigt model which includes another

spring and a damper connected in series the Maxwell model. The Voigt model does not

accommodate the inertial effect of the fluid present in the system and it becomes invalid.

The Maxwell-Voigt model is used instead. Voigt model and Maxell-Voigt model are

good enough when it comes to capturing the characteristics of isolators that are used in
 18

applications that possesses small range of frequency and over small bands of

displacements amplitudes. On the other hand, when it comes to applications where the

isolators are used over a large operating frequency range, the above models may not be

sufficient to capture the hysteretic behavior. In this case the hysteretic model based on

Bouc-Wen model is used. Bouc-Wen model is a nonlinear model that has the capability

of capturing the time dependency by introducing an additional state variable.

Ye and Wang (2007) conducted a study to estimate the Bouc-Wen model

parameters. The proposed approach use particle swarm optimization (PSO) which is

based on the movement and intelligence of swarms. The results of the PSO method are

compared to the genetic algorithms (GAs) in terms of parameter accuracy. It is shown

that higher quality solution and better computational efficiency can be achieved by using

the PSO method.

Ikhouane, et al. (2006) focus on the fact that even if there is a good approximation

of the true hysteresis modeled using the Bouc-Wen model, it may not keep significant

physical properties which are inherent in the real data. The work in this paper presents a

characterization of the different classes of Bouc-Wen models in terms of their bounded

input bounded output stability and as a result reproducing the physical properties inherent

in true systems that have been modeled.

2.4 Load Estimation

Generally speaking, in order to accurately design any component, good estimate

of the loads acting on the component is vital. The stresses induced in a component are a

function of the loads applied. The accuracy of estimating the loads applied to the engine

mount plays an important role in designing the mounting system and its components (i.e.
 19

stiffness and damping). The major problem that arises is measuring the loads. The

simplest way of estimating the loads is by a direct measure using load cells. In some

cases, inserting load cells is almost impossible due the nature of the structure. Another

approach which is recently attracting attention is treating the structure as a load

transducer. In this technique, the measured strains on some parts of the structure can

provide a history of the loads acting on it

Masroor and Zachary (1990) proposed a procedure to select the location of strain

gauges on a structure. The procedure is valid for linear elastic static problems. It can

accommodate both isotropic and nonisotropic materials. The procedure involves applying

a unit load each time and collecting the corresponding strains. This will produces a

matrix that contains the strain information at the candidate locations of the strain gauges.

The selection of the final location of the strain gauges is determined by the best

approximate solution (BAS) that minimizes the sum of the squared errors.

Wickham et al. (1995) presented a computational tool that uses the D-optimal

design technique to find the location and orientation of the strain gauges. The tool insures

a precise location of the selected strain gauges. This in return will insure an accurate load

recovery.

Dhingra and Hunter (2003) proposed a technique that considers the whole

structure as a load transducer. The technique is valid for both 2 dimensional and 3

dimensional structures as well. The procedure delivers the location as well as the

orientation of the strain gauges to be used through the help of finite element software.

The selection of the strain gauge locations and orientation is done by using optimization
 20

technique. This is achieved by using the D-optimal design procedure which maximizes

the determinant of the matrix (ATA).

While load estimation using strain gauges mounted on the structure is a promising

approach, it is not used herein due to resources limitations. Instead, the approach adopted

estimates external loads transmitted through the mount system by monitoring the road

profile in contact with the tire patch.

2.5 Shape Optimization

Once the mount is designed, i.e. the stiffness and damping values of the mount are

known, the next step is to translate these numerical values into physical mount. This

involves determining geometrical dimensions of the mount such that it have desired

stiffness and damping characteristics

Kim J. and Kim Heon (1997) conducted a study on bush (shear) type engine mount

that is used frequently in the auto industry in order to come up with the optimum

geometrical shape of the mount. The study is performed by utilizing nonlinear finite

element commercial software. The main objective is to minimize the difference between

a set of target stiffness values in the three principal directions obtained from dynamic

analysis with the stiffness values in the same directions generated from the geometry of

the mount. In this process, a set of variables that fully describes the mount are used as the

design vector to be determined from the outcome of the optimization problem.

Ali, et al (2010) conducted a study reviewing the need to different types of

constitutive modes for rubber like materials. Modeling of these Elastomers depends on

the strain energy function. The selection of the proper rubber elastic material is essential.
 21

The stress-strain response is required in order to define the hyperelastic material

behavior.

Scharnhorst and Pain (1978) utilized the Reissner type variational principle to

formulate a mixed finite element model of finite strain analysis for Mooney-Rivlin like

materials. They have adopted an incremental and stationary Lagrangian formulation. The

variables consist of incremental displacements and incremental hydrostatic and

distortional stresses. Four node quadrilateral plane strain elements were used in this work

to analyze the inflation of an infinitely long thick-walled cylinder subjected to internal

pressure.

Swanson (1985) noted that a certain type of problems in which the finite

compressibility of high elongation rubber like materials influence the stress distribution,

as a result must be taken into consideration. They addressed the problem by introducing a

new rubber elasticity model with finite compressibility and improved material

representation.

2.6 Summary

A fair bit of work has been done in the area of mount system design and isolation.

The primary goal is to achieve an enhanced performance when it comes to isolation. This

is done by better understanding the isolator and its components. Mechanical snubbing is a

major aspect that needs to be taken into consideration when designing any mounting

system. This dissertation will address the snubbing problem in chapter 3. This will be

done in the context of several alternate mount models.

In addition, the dissertation will also look into at the external loads transmitted

from road bumps through the mount system. Two criteria are used for designing the
 22

mounting system namely; minimizing the force transmitted through the mount system or

designs the system to decouple the vibration modes. Finally, the geometrical shape of the

engine mount will be determined at different engine operating speeds.

 23

3 Chapter 3 – Mount Characterization and Determination of Mount

Parameters

This chapter discusses the basic idea of mount characterization. The process starts

with load estimation. The loads are transmitted to the frame through the tire patch.

Multiple models for the engine mounts are developed in this chapter. These models vary

in complexity from a simple Voigt model to a complex Voigt model that incorporates

nonlinear stiffness, mechanical snubbing and a Bouc-Wen element to capture mount

hysteresis. The mount parameters are then identified by minimizing the difference

between the theoretical transmitted forces and the experimentally measured forces. All

the necessary equations of motion and the mathematical equations for the theoretical

transmitted forces are developed in this chapter. An optimization problem is formulated

to help determine the mount parameters.

3.1 Load Estimation

A mounting system is mainly used to minimize vibrations and shaking forces

from the engine from being passed on to the frame, and eventually to the passengers.

Also, the mount system might serve another purpose such as minimizing the forces and

vibrations due to road bumps from being transmitted to the powertrain. As a result, an

important issue when designing a mounting system is figuring out the forces passing

through the mount system that needs to be minimized.

The problem occurs when trying to estimate the forces being transmitted through

the whole system. The forces can be measured directly by using load cells which might

not be easy due to the nature of the structure. One method that can be used in the case of
 24

loads being transmitted from the road bumps through the tire patch is modeling the input

force as shown in Eq. 3.1.

𝐹𝐹𝑦𝑦 (𝑡𝑡) = 𝑘𝑘 𝑥𝑥(𝑡𝑡) + 𝑐𝑐 𝑥𝑥̇ (𝑡𝑡) (3.1)

In Eq. (3.1) 𝐹𝐹𝑦𝑦 (𝑡𝑡) is the forces transmitted in the y direction through the tire patch due to

the displacement 𝑥𝑥(𝑡𝑡) and velocity 𝑥𝑥̇ (𝑡𝑡) caused by the change of the road profile as

shown in Fig. 3.1. 𝑘𝑘 and 𝑐𝑐 are the stiffness and damping of the wheels and the suspension

in the y direction.

An alternate approach for estimating the forces acting on the system is by treating

the structure as a load transducer and by measuring the strains at some previously

determined locations. In order to find the most appropriate location and orientation of the

strain gauges to place on the structure, there is a need to perform finite element analysis.

By knowing the proper location and orientation of the strain gauges, and the use of the

principle of superposition, the loads acting on the structure can be determined. This

procedure however is not considered in this work.

Figure 3.1: Schematic Diagram Showing the Tire Patch

 25

3.2 Elastomeric Mount

Engine mounts that are used in automotive industry are primarily made of rubber.

The rubber stiffness is categorized as either static or dynamic. In motorcycle or

automotive industry, the sag of the powertrain due to the static weight is described using

the static stiffness. On the other hand, the dynamic stiffness is used to determine the

vibration isolation as a result of the application of a harmonic load. The dynamic stiffness

varies with the amplitude and frequency of the applied load or the applied displacement.

The relation that governs the static and the dynamic stiffness is as follows:

𝐾𝐾𝑑𝑑𝑑𝑑𝑑𝑑 = 𝜂𝜂 𝐾𝐾𝑠𝑠𝑠𝑠 (3.2)

In Eq. (3.2), 𝜂𝜂 is the dynamic to static coefficient which is always greater than 1. Kst is

the static stiffness and Kdyn is the dynamic stiffness. The dynamic to static coefficient

varies with the input frequency leading to a higher coefficient with higher frequencies

and as a result a higher dynamic stiffness.

The complex stiffness for an elastomer that is subjected to a sinusoidal

displacement 𝑥𝑥𝑖𝑖 (𝑡𝑡) with an output force 𝑓𝑓𝑜𝑜 (𝑡𝑡) is the output force to the input

displacement described as follows:

∗
𝐹𝐹𝑜𝑜∗ 𝐹𝐹𝑜𝑜 𝐹𝐹𝑜𝑜
𝐾𝐾 = = 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 + 𝑗𝑗 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 (3.3)
𝑋𝑋𝑖𝑖 𝑋𝑋𝑖𝑖 𝑋𝑋𝑖𝑖

The displacement 𝑥𝑥𝑖𝑖 (𝑡𝑡) = 𝑋𝑋𝑖𝑖 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑗 and the input force 𝑓𝑓𝑜𝑜 (𝑡𝑡) = 𝐹𝐹𝑜𝑜 𝑒𝑒 𝑗𝑗 (𝜔𝜔𝜔𝜔 +𝛿𝛿) = 𝐹𝐹𝑜𝑜∗ 𝑒𝑒 𝑗𝑗𝑗𝑗 ,

where 𝐹𝐹𝑜𝑜∗ = 𝐹𝐹𝑜𝑜 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑗 .

In Eq. (3.3), 𝑋𝑋𝑖𝑖 is the peak displacement amplitude, 𝐹𝐹𝑜𝑜 is the peak force

amplitude, 𝛿𝛿 is the phase angle between the input displacement and the output force and

𝜔𝜔 is the input frequency. Eq. (3.3) can be generally expressed as follows:

 26

𝐾𝐾 ∗ = 𝐾𝐾 ′ + 𝑗𝑗𝐾𝐾 ′′ (3.4)

where,

𝐹𝐹𝑜𝑜 𝐹𝐹𝑜𝑜
𝐾𝐾 ′ = 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 and 𝐾𝐾 ′′ 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 (3.5)
𝑋𝑋𝑖𝑖 𝑋𝑋𝑖𝑖

The elastomer dynamic stiffness 𝐾𝐾𝑑𝑑𝑑𝑑𝑑𝑑 is the magnitude of 𝐾𝐾 ∗ .

𝐹𝐹𝑜𝑜
𝐾𝐾𝑑𝑑𝑑𝑑𝑑𝑑 = |𝐾𝐾 ∗ | = �(𝐾𝐾 ′ )2 + (𝐾𝐾 ′′ )2 = (3.6)
𝑋𝑋𝑖𝑖

The loss factor is defined as follows:

𝐾𝐾 ′′
𝛽𝛽 = 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 = (3.7)
𝐾𝐾 ′

The loss factor 𝛽𝛽 is used to determine the damping or the hysteresis of the engine

mount. Both the dynamic stiffness and the loss factor are critical parameters in the

modeling of the engine mounts.

The engine mount is typically represented as three mutually orthogonal

translational springs about the center of elasticity shown in Fig. 3.2. In Fig. 3.2, the

coordinate system of the mount is also shown. It is assumed that the mount is attached to

the rigid body by means of ball joints. This implies that the resilient element is incapable

of applying a moment to the body it is attached to.

 27

Figure 3.2: Tri-axial Engine Mount Model

3.3 Mount Modeling

The main component in designing a mounting system is the design of the mount

itself. Identifying the mount parameters such as the mount stiffness and damping is a

crucial step in the process of designing an appropriate mounting system. There are four

candidate models that are used to represent the mounting system, all of which are

assumed to be elastomeric isolators. The first model is the Voigt model shown in Fig. 3.3.

This model is formulated using a spring and damper that are connected in parallel to a

supporting mass. The second model is the Maxwell-Voigt model shown in Fig. 3.4. This

model is formulated like the Voigt model and has an additional spring and a damper

connected in series. The third model is the Voigt model with a Bouc-Wen element as

shown in Fig. 3.5. This model is formulated like the first model with the addition of a

Bouc-Wen element to capture mount hysteresis. The fourth Model is the Voigt Model

with a Bouc-Wen element and nonlinear stiffness as shown in Fig. 3.6. This model is the

same as the third model with one exception; the snubbing stiffness is modeled as a

nonlinear spring.
 28

All of the models presented next will have an element that represents mechanical

snubbing. A mechanical snubber is used in mounts to absorb large amount of energy

within small displacement amplitudes, and is modeled as a spring with linear stiffness for

all of the models except for the last model where snubbing is modeled using spring that

possesses a nonlinear stiffness. Mechanical snubbers are very important when it comes to

designing an isolation system. It is used as a device to limit the motion of the mounting

system when it undergoes overloading conditions.

3.3.1 Model 1 - Voigt Model

Fig. 3.3 shows the configuration of model 1. The model consists of a single

degree of freedom system where the spring and damper are represented by the stiffness

𝑘𝑘1 and damping coefficient 𝑏𝑏1 . The snubbing effect is taken into account by adding

additional two linear springs 𝑘𝑘2 and 𝑘𝑘3 that will be engaged when the displacement

amplitude 𝑥𝑥 of the isolated mass exceeds the snubbing gap 𝑥𝑥𝑜𝑜 . The equations of motion

for model 1 are as follows:

𝑚𝑚𝑥𝑥̈ + 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 = 𝑓𝑓, for |𝑥𝑥| < 𝑥𝑥𝑜𝑜

𝑚𝑚𝑥𝑥̈ + 𝑏𝑏1 𝑥𝑥̇ + (𝑘𝑘1 + 𝑘𝑘2 )𝑥𝑥 = 𝑓𝑓 + 𝑘𝑘2 𝑥𝑥𝑜𝑜 , for 𝑥𝑥 ≥ 𝑥𝑥𝑜𝑜 (3.8)

𝑚𝑚𝑥𝑥̈ + 𝑏𝑏1 𝑥𝑥̇ + (𝑘𝑘1 + 𝑘𝑘3 )𝑥𝑥 = 𝑓𝑓 − 𝑘𝑘3 𝑥𝑥𝑜𝑜 , for 𝑥𝑥 ≤ −𝑥𝑥𝑜𝑜

In the equation above, 𝑚𝑚 represents the mass of the isolated system and 𝑓𝑓 is the

excitation force acting on the system. The system of 2nd order linear differential equations

in Eq. (3.8) can be converted into a system 1st order linear differential equations as

follows:

𝑥𝑥̇ 1 0 1 𝑥𝑥1 0
� � = �−𝑘𝑘1 −𝑏𝑏1 � � � + � 𝑓𝑓 � (3.9)
𝑥𝑥̇ 2 𝑥𝑥2
𝑚𝑚 𝑚𝑚 𝑚𝑚
 29

𝑥𝑥̇ 1 0 1 𝑥𝑥1 0
� � = �−(𝑘𝑘1 + 𝑘𝑘2 ) −𝑏𝑏� � � + � 𝑓𝑓 𝑘𝑘2 𝑥𝑥𝑜𝑜 � (3.10)
𝑥𝑥̇ 2 𝑥𝑥2 +
𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚

𝑥𝑥̇ 1 0 1 𝑥𝑥1 0
� � = �−(𝑘𝑘1 + 𝑘𝑘3 ) −𝑏𝑏� � � + � 𝑓𝑓 𝑘𝑘3 𝑥𝑥𝑜𝑜 � (3.11)
𝑥𝑥̇ 2 𝑥𝑥2 −
𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚

In the above equations, 𝑥𝑥1 = 𝑥𝑥 and 𝑥𝑥2 = 𝑥𝑥̇ . Eq. (3.9) is the governing equation

of motion for the system shown in Fig. 3.3 when |𝑥𝑥| < 𝑥𝑥𝑜𝑜 . Eq. (3.10) is the governing

equation of motion when 𝑥𝑥 ≥ 𝑥𝑥𝑜𝑜 and Eq. (3.11) is the governing equation of motion

when 𝑥𝑥 ≤ −𝑥𝑥𝑜𝑜 . To accommodate varying stiffness in different directions of motion, the

snubbing stiffness 𝑘𝑘2 and 𝑘𝑘3 is assumed to be asymmetrical. This assumption can be

relaxed for symmetric systems by making the snubbing stiffness equal for both motion

directions.

𝑘𝑘2

x f
𝑥𝑥𝑜𝑜

𝑥𝑥𝑜𝑜

𝑘𝑘1
𝑏𝑏1
𝑘𝑘3

Figure 3.3: Voigt Model

 30

3.3.2 Model 2 - Maxwell-Voigt Model

Fig. 3.4 shows the configuration of model 2. This model consists of a single

degree of freedom system where the spring and damper are represented by stiffness 𝑘𝑘1

and damping 𝑏𝑏1 just like in model 1. However, there is an additional spring with stiffness

𝑘𝑘 and a damper with damping coefficient 𝑏𝑏 connected in series in reference to Maxwell-

Voigt model. The snubbing effect is modeled just like the one presented in model 1. The

equations of motion for the system shown in Fig. 3.4 are as follows:

𝑚𝑚 0 𝑥𝑥̈ 𝑏𝑏1 0 𝑥𝑥̇ (𝑘𝑘1 + 𝑘𝑘) −𝑘𝑘 𝑥𝑥 𝑓𝑓

� �� � + � �� � + � �� � = � � (3.12)
0 0 𝑦𝑦̈ 0 𝑏𝑏 𝑦𝑦̇ −𝑘𝑘 𝑘𝑘 𝑦𝑦 0

𝑚𝑚 0 𝑥𝑥̈ 𝑏𝑏1 0 𝑥𝑥̇ (𝑘𝑘1 + 𝑘𝑘2 + 𝑘𝑘) −𝑘𝑘 𝑥𝑥 𝑓𝑓 + 𝑘𝑘2 𝑥𝑥𝑜𝑜

� �� � + � �� � + � �� � = � � (3.13)
0 0 𝑦𝑦̈ 0 𝑏𝑏 𝑦𝑦̇ −𝑘𝑘 𝑘𝑘 𝑦𝑦 0

𝑚𝑚 0 𝑥𝑥̈ 𝑏𝑏1 0 𝑥𝑥̇ (𝑘𝑘1 + 𝑘𝑘3 + 𝑘𝑘) −𝑘𝑘 𝑥𝑥 𝑓𝑓 − 𝑘𝑘3 𝑥𝑥𝑜𝑜

� �� �+ � �� �+ � �� � = � � (3.14)
0 0 𝑦𝑦̈ 0 𝑏𝑏 𝑦𝑦̇ −𝑘𝑘 𝑘𝑘 𝑦𝑦 0

In Eq. (3.12) through Eq. (3.14), 𝑘𝑘 and 𝑏𝑏 represents the stiffness and damping of

the additional spring and the damper added to the Maxwell model as shown in Fig. 3.4.

Eq. (3.12) through Eq. (3.14) can be expressed as system of 1st order differential equation

as follows:

𝑥𝑥̇ 1 0 1 0 𝑥𝑥1 0
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
⎢ ⎥ ⎢−(𝑘𝑘 + 𝑘𝑘1 ) 𝑏𝑏1 𝑘𝑘 ⎥
⎢𝑥𝑥 ⎥ ⎢ 𝑓𝑓 ⎥
⎢ −
⎢𝑥𝑥̇ 2 ⎥ = ⎢ 𝑚𝑚 𝑚𝑚 𝑚𝑚 ⎥⎥ ⎢ 2 ⎥ + ⎢𝑚𝑚⎥ (3.15)
⎢ ⎥ ⎢ 𝑘𝑘 𝑘𝑘 ⎢ ⎥ ⎢ ⎥
⎣ 𝑦𝑦̇ ⎦ ⎣ 0 − ⎥⎦ ⎣ 𝑦𝑦 ⎦ ⎣ 0 ⎦
𝑏𝑏 𝑏𝑏

𝑥𝑥̇ 1 0 1 0 𝑥𝑥1 0
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
⎢ ⎥ ⎢−(𝑘𝑘 + 𝑘𝑘1 + 𝑘𝑘2 ) 𝑏𝑏1 𝑘𝑘 ⎥
⎢𝑥𝑥 ⎥ ⎢ 𝑓𝑓 𝑘𝑘2 ⎥
⎢ −
⎢𝑥𝑥̇ 2 ⎥ = ⎢ 𝑚𝑚 𝑚𝑚 𝑚𝑚 ⎥⎥ ⎢ 2 ⎥ + ⎢𝑚𝑚 + 𝑚𝑚 𝑥𝑥𝑜𝑜 ⎥ (3.16)
⎢ ⎥ ⎢ 𝑘𝑘 𝑘𝑘 ⎢ ⎥ ⎢ ⎥
⎣ 𝑦𝑦̇ ⎦ ⎣ 0 − ⎥⎦ ⎣ 𝑦𝑦 ⎦ ⎣ 0 ⎦
𝑏𝑏 𝑏𝑏
 31

𝑥𝑥̇ 1 0 1 0 𝑥𝑥1 0
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
⎢ ⎥ ⎢−(𝑘𝑘 + 𝑘𝑘1 + 𝑘𝑘3 ) 𝑏𝑏1 𝑘𝑘 ⎥
⎢𝑥𝑥 ⎥ ⎢ 𝑓𝑓 𝑘𝑘3 ⎥
⎢ −
⎢𝑥𝑥̇ 2 ⎥ = ⎢ 𝑚𝑚 𝑚𝑚 𝑚𝑚 ⎥⎥ ⎢ 2 ⎥ + ⎢𝑚𝑚 − 𝑚𝑚 𝑥𝑥𝑜𝑜 ⎥ (3.17)
⎢ ⎥ ⎢ 𝑘𝑘 𝑘𝑘 ⎢ ⎥ ⎢ ⎥
⎣ 𝑦𝑦̇ ⎦ ⎣ 0 − ⎥⎦ ⎣ 𝑦𝑦 ⎦ ⎣ 0 ⎦
𝑏𝑏 𝑏𝑏

𝑘𝑘2

x f
𝑥𝑥𝑜𝑜

𝑥𝑥𝑜𝑜
y
k
𝑘𝑘1
𝑏𝑏1
𝑘𝑘3
b

Figure 3.4: Maxwell-Voigt Model

3.3.3 Model 3 - Voigt Model with Bouc-Wen Element

Fig. 3.5 shows the configuration of model 3. This model is similar to model 1. In

order to capture the hysteretic behavior over large range of operating frequencies, a

Bouc-Wen element is added (Ikhouane, 2006). The Bouc-Wen element is a nonlinear

element that is added to the model to capture the time dependence by adding the time

dependent parameter (𝑧𝑧). 𝛼𝛼, 𝛽𝛽, 𝛾𝛾, 𝐴𝐴 𝑎𝑎𝑎𝑎𝑎𝑎 𝑛𝑛 are the set of Bouc-Wen element parameters
 32

that need to be defined. The model that incorporates the Bouc-Wen element is shown in

Fig. 3.5 and can be expressed as a system 1st order differential equations as follows:

𝑥𝑥̇ 1 = 𝑥𝑥2

𝑘𝑘1 𝑏𝑏1 𝛼𝛼 𝑓𝑓
𝑥𝑥̇ 2 = − 𝑥𝑥1 − 𝑥𝑥2 − 𝑧𝑧 + (3.18)
𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚

𝑧𝑧̇ = −𝛾𝛾|𝑥𝑥2 |𝑧𝑧|𝑧𝑧|𝑛𝑛−1 − 𝛽𝛽𝑥𝑥2 |𝑧𝑧|𝑛𝑛 + 𝐴𝐴𝑥𝑥2

The nonlinear system shown above in Eq. (3.18) holds when there is no snubbing

effect i.e. |𝑥𝑥| < 𝑥𝑥𝑜𝑜 . 𝑥𝑥1 = 𝑥𝑥 and 𝑥𝑥2 = 𝑥𝑥̇ . 𝛼𝛼, 𝛽𝛽, 𝛾𝛾, 𝐴𝐴 and 𝑛𝑛 are constants referred to as

Bouc-Wen parameters. z is the time varying constant introduced by Bouc-Wen element.

When 𝑥𝑥 ≥ 𝑥𝑥𝑜𝑜 , the equations of motion (EOM) are as follows:

𝑥𝑥̇ 1 = 𝑥𝑥2

(𝑘𝑘1 + 𝑘𝑘2 ) 𝑏𝑏1 𝛼𝛼 𝑓𝑓 𝑘𝑘2

𝑥𝑥̇ 2 = − 𝑥𝑥1 − 𝑥𝑥2 − 𝑧𝑧 + + 𝑥𝑥 (3.19)
𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑜𝑜

𝑧𝑧̇ = −𝛾𝛾|𝑥𝑥2 |𝑧𝑧|𝑧𝑧|𝑛𝑛−1 − 𝛽𝛽𝑥𝑥2 |𝑧𝑧|𝑛𝑛 + 𝐴𝐴𝑥𝑥2

and finally when 𝑥𝑥 ≤ −𝑥𝑥𝑜𝑜 , the equations of motion (EOM) are as follows:

𝑥𝑥̇ 1 = 𝑥𝑥2

(𝑘𝑘1 + 𝑘𝑘3 ) 𝑏𝑏1 𝛼𝛼 𝑓𝑓 𝑘𝑘3

𝑥𝑥̇ 2 = − 𝑥𝑥1 − 𝑥𝑥2 − 𝑧𝑧 + − 𝑥𝑥 (3.20)
𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑜𝑜

𝑧𝑧̇ = −𝛾𝛾|𝑥𝑥2 |𝑧𝑧|𝑧𝑧|𝑛𝑛−1 − 𝛽𝛽𝑥𝑥2 |𝑧𝑧|𝑛𝑛 + 𝐴𝐴𝑥𝑥2

 33

𝑘𝑘2

x f
𝑥𝑥𝑜𝑜

𝑥𝑥𝑜𝑜

𝑘𝑘1
𝑏𝑏1
𝑘𝑘3

Figure 3.5: Voigt Model with Bouc-Wen Element

3.3.4 Model 4 - Voigt Model with Bouc-Wen Element and Nonlinear

Stiffness

Fig. 3.6 shows the configuration of model 4. This model is modeled like model 3.

However, in all of the models mentioned above, the snubbing is represented as a linear

spring. In model 4, snubbing is represented by a nonlinear stiffness to capture the

progressive stiffening behavior when the snubber is engaged. The governing EOM for

model 4 are the same as defined in Eq. (3.18) when the snubber is not engaged i.e.

|𝑥𝑥| < 𝑥𝑥𝑜𝑜 . When 𝑥𝑥 ≥ 𝑥𝑥𝑜𝑜 , the EOM for the model are as follows:
 34

𝑥𝑥̇ 1 = 𝑥𝑥2

𝑘𝑘1 𝑘𝑘2 𝑘𝑘2 𝑏𝑏1 𝛼𝛼 𝑓𝑓

𝑥𝑥̇ 2 = − 𝑥𝑥1 − (𝑥𝑥1 − 𝑥𝑥𝑜𝑜 ) − (𝑥𝑥1 − 𝑥𝑥𝑜𝑜 )3 − 𝑥𝑥2 − 𝑧𝑧 + (3.21)
𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚

𝑧𝑧̇ = −𝛾𝛾|𝑥𝑥2 |𝑧𝑧|𝑧𝑧|𝑛𝑛−1 − 𝛽𝛽𝑥𝑥2 |𝑧𝑧|𝑛𝑛 + 𝐴𝐴𝑥𝑥2

In Eq. (3.21), k2 represents the snubber stiffness which is modeled as a cubic

nonlinear relationship instead of the linear snubber stiffness used in the previous models.

The EOM of for 𝑥𝑥 ≤ 𝑥𝑥𝑜𝑜 are as follows:

𝑥𝑥̇ 1 = 𝑥𝑥2

𝑘𝑘1 𝑘𝑘3 𝑘𝑘3 𝑏𝑏1 𝛼𝛼 𝑓𝑓

𝑥𝑥̇ 2 = − 𝑥𝑥1 − (𝑥𝑥1 + 𝑥𝑥𝑜𝑜 ) − (𝑥𝑥1 + 𝑥𝑥𝑜𝑜 )3 − 𝑥𝑥2 − 𝑧𝑧 + (3.22)
𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚

𝑧𝑧̇ = −𝛾𝛾|𝑥𝑥2 |𝑧𝑧|𝑧𝑧|𝑛𝑛−1 − 𝛽𝛽𝑥𝑥2 |𝑧𝑧|𝑛𝑛 + 𝐴𝐴𝑥𝑥2

𝑘𝑘2

x f
𝑥𝑥𝑜𝑜

𝑥𝑥𝑜𝑜

𝑘𝑘1
𝑏𝑏1
𝑘𝑘3

Figure 3.6: Voigt Model with Bouc-Wen Element and Nonlinear Stiffness
 35

3.4 Parameter Identification

Parameter identification is used next to find the variables associated with each of

the models presented in sections 3.3.1 to 3.3.4. An optimization procedure is used to

determine the variables in each of the presented models. The solution technique involves

finding the difference between the transmitted force computed from the theoretical

models and the measured force. A time history of the measured forces transmitted to the

base from a known excitation is used for model reconciliation.

The objective function for the parameter identification is defined as:

‖𝑓𝑓𝑡𝑡 − 𝑓𝑓𝑚𝑚 ‖ (3.23)

In Eq. (3.23), 𝑓𝑓𝑡𝑡 is the time history of the transmitted force to the base calculated from the

theoretical model and 𝑓𝑓𝑚𝑚 is the time history of the measured force for a specified input.

This solution technique works by minimizing the norm of the difference between the time

history of the two forces, which results in finding the relevant parameters of the

corresponding models. In the optimization problem, the only constraints are the side

constraints which provide a limit for the design variables except for the models in section

3.3.3 and 3.3.4, where the presence of the constraints is necessary to limit the Bouc-Wen

model parameters in order to insure bounded input bounded output response (Ikhouane,

2006). The Sequential Quadratic Programming (SQP) algorithm is used to the

optimization problem. The ‘fmincon’ function in MATLAB® optimization toolbox is

used to minimize the function in Eq. (3.23).

The force transmitted to the base for the Voigt model, defined in Fig. 3.3, is

expressed as follows:
 36

𝑓𝑓𝑡𝑡 = 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥, for |𝑥𝑥| < 𝑥𝑥𝑜𝑜

𝑓𝑓𝑡𝑡 = 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝑘𝑘2 (𝑥𝑥 − 𝑥𝑥𝑜𝑜 ), for 𝑥𝑥 ≥ 𝑥𝑥𝑜𝑜

𝑓𝑓𝑡𝑡 = 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝑘𝑘3 (𝑥𝑥 + 𝑥𝑥𝑜𝑜 ), for 𝑥𝑥 ≤ −𝑥𝑥𝑜𝑜 (3.24)

In Eq. (3.24), 𝑘𝑘1 , 𝑘𝑘2 , 𝑘𝑘3 and 𝑏𝑏1 are the system parameters that need to be determined from

the parameter identification for the Voigt model.

The force transmitted to the base for the Maxwell-Voigt model, defined in Fig. 3.4,

is expressed as follows:

𝑓𝑓𝑡𝑡 = 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝑏𝑏𝑦𝑦̇

= 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝑘𝑘(𝑥𝑥 − 𝑦𝑦), for |𝑥𝑥| < 𝑥𝑥𝑜𝑜

𝑓𝑓𝑡𝑡 = 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝑏𝑏𝑦𝑦̇ + 𝑘𝑘2 (𝑥𝑥 − 𝑥𝑥𝑜𝑜 )

= 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝑘𝑘(𝑥𝑥 − 𝑦𝑦) + 𝑘𝑘2 (𝑥𝑥 − 𝑥𝑥𝑜𝑜 ), for 𝑥𝑥 ≥ 𝑥𝑥𝑜𝑜 (3.25)

𝑓𝑓𝑡𝑡 = 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝑏𝑏𝑦𝑦̇ + 𝑘𝑘3 (𝑥𝑥 + 𝑥𝑥𝑜𝑜 )

= 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝑘𝑘(𝑥𝑥 − 𝑦𝑦) + 𝑘𝑘3 (𝑥𝑥 + 𝑥𝑥𝑜𝑜 ), for 𝑥𝑥 ≤ −𝑥𝑥𝑜𝑜

In Eq. (3.25), 𝑘𝑘1 , 𝑘𝑘2 , 𝑘𝑘3 , 𝑘𝑘, 𝑏𝑏1 and 𝑏𝑏 are the system parameters that need to determined

from the parameter identification for the Maxwell-Voigt model.

The force transmitted to the base for the Voigt model with Bouc-Wen element,

defined in Fig. 3.5 is expressed as follows:

𝑓𝑓𝑡𝑡 = 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝛼𝛼𝛼𝛼, for |𝑥𝑥| < 𝑥𝑥𝑜𝑜

𝑓𝑓𝑡𝑡 = 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝛼𝛼𝛼𝛼 + 𝑘𝑘2 (𝑥𝑥 − 𝑥𝑥𝑜𝑜 ), for 𝑥𝑥 ≥ 𝑥𝑥𝑜𝑜 (3.26)

𝑓𝑓𝑡𝑡 = 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝛼𝛼𝛼𝛼 + 𝑘𝑘3 (𝑥𝑥 + 𝑥𝑥𝑜𝑜 ), for 𝑥𝑥 ≤ −𝑥𝑥𝑜𝑜

In Eq. (3.26), 𝑘𝑘1 , 𝑘𝑘2 , 𝑘𝑘3 , 𝑏𝑏1 and 𝛼𝛼 need to be determined. 𝑧𝑧 in the equation above is a time

varying variable that comes as a result of using Bouc-Wen element which depends on the
 37

parameters 𝛽𝛽, 𝛾𝛾, 𝑛𝑛 and 𝐴𝐴 which need to be determined also for the Voigt model with

Bouc-Wen element.

The force transmitted to the base for the Voigt model with Bouc-Wen element

and nonlinear stiffness, defined in Fig. 3.6, is expressed as follows:

𝑓𝑓𝑡𝑡 = 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝛼𝛼𝛼𝛼, for |𝑥𝑥| < 𝑥𝑥𝑜𝑜

𝑓𝑓𝑡𝑡 = 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝛼𝛼𝛼𝛼 + 𝑘𝑘2 (𝑥𝑥 − 𝑥𝑥𝑜𝑜 ) + 𝑘𝑘2 (𝑥𝑥 − 𝑥𝑥𝑜𝑜 )3 , for 𝑥𝑥 ≥ 𝑥𝑥𝑜𝑜 (3.27)

𝑓𝑓𝑡𝑡 = 𝑏𝑏1 𝑥𝑥̇ + 𝑘𝑘1 𝑥𝑥 + 𝛼𝛼𝛼𝛼 + 𝑘𝑘3 (𝑥𝑥 + 𝑥𝑥𝑜𝑜 ) + 𝑘𝑘3 (𝑥𝑥 + 𝑥𝑥𝑜𝑜 )3 , for 𝑥𝑥 ≤ −𝑥𝑥𝑜𝑜

In Eq. (3.27), 𝑘𝑘1 , 𝑘𝑘2 , 𝑘𝑘3 , 𝑏𝑏1 and 𝛼𝛼 need to be determined. 𝑧𝑧 which depends on

𝛽𝛽, 𝛾𝛾, 𝑛𝑛 and 𝐴𝐴 are also need to be determined for the Voigt model with Bouc-Wen element

and nonlinear stiffness.

3.4.1 Numerical example

To illustrate the parameter identification procedure, consider the following test

situation. Experimental test results obtained from an elastomeric engine mount are used

to find the mount parameters. The experimental force-displacement data was furnished by

Dr. Kaul. The force-displacement data collected from the experiment is shown in Fig.

3.7. The data computed for the four models is shown in Table 3.1 and the results for the

force-displacement relationship for the four models are shown in Figs. 3.8, 3.9, 3.10 and

3.11 respectively.
 38

Measured Force - Displacement Data

1500

1000

500

0
Force (N)

-500

-1000

-1500

-2000
-8 -6 -4 -2 0 2 4 6 8
Displacement (mm)

Figure 3.7: Measured Force – Displacement Curve.

Estimated Curve - Voigt Model

1500

1000

500

0
Force (N)

-500

-1000

-1500

-2000
-8 -6 -4 -2 0 2 4 6 8
Displacement (mm)

Figure 3.8: Force Displacement Curve (Voigt Model)

 39

Estimated Curve - Maxwell-Voigt Model

1500

1000

500

0
Force (N)

-500

-1000

-1500

-2000
-8 -6 -4 -2 0 2 4 6 8
Displacement (mm)

Figure 3.9: Force Displacement Curve (Maxwell-Voigt Model)

Estimated Curve - Voigt Model with Bouc-Wen Element

1500

1000

500

0
Force (N)

-500

-1000

-1500

-2000
-8 -6 -4 -2 0 2 4 6 8
Displacement (mm)

Figure 3.10: Force Displacement Curve (Voigt Model with Bouc-Wen Element)
 40

Estimated Curve - Voigt Model with Bouc-Wen Element and Nonlinear Stiffness
1500

1000

500

0
Force (N)

-500

-1000

-1500

-2000
-8 -6 -4 -2 0 2 4 6 8
Displacement (mm)

Figure 3.11: Force Displacement Curve (Voigt Model with Bouc-Wen Element and
Nonlinear Stiffness)
 41

Table 3.1: Computed Parameters for the Four Models

Stiffness Parameters Damping Parameters Bouc-Wen Optimized

(N/mm) (N-s/mm) Parameters F

𝑘𝑘1 = 116.31

Model 1 𝑘𝑘2 = 314.83 𝑏𝑏1 = 1.38 5.44E+03

𝑘𝑘3 = 746.35

𝑘𝑘1 = 112.35

Model 2 𝑘𝑘2 = 299.29 𝑏𝑏1 = 0.001

𝑘𝑘3 = 732.14 𝑏𝑏 = 2.78

𝑘𝑘 = 45.18 5.82E+03

𝛼𝛼 = 0.01

𝑘𝑘1 = 116.31 𝛽𝛽 = 0.48

Model 3 𝑘𝑘2 = 314.83 𝑏𝑏1 = 1.37 𝛾𝛾 = 1

𝑘𝑘3 = 746.35 𝑛𝑛 = 1.65

𝐴𝐴 = 1.1 5.44E+03

𝛼𝛼 = 0.01

𝑘𝑘1 = 118.76 𝛽𝛽 = 0.5

Model 4 𝑘𝑘2 = 154.31 𝑏𝑏1 = 1.36 𝛾𝛾 = 0.94

𝑘𝑘3 = 429.62 𝑛𝑛 = 1.42

𝐴𝐴 = 0.83 4.81E+03
 42

The parameters for the four models, based on the formulation presented in sections

3.3 and 3.4, are presented in Table 3.1. As can be seen from Table 3.1, the value for the

isolation stiffness variable 𝑘𝑘1 , varies by 5% between the four models. The same can be

said about 𝑏𝑏1 , except for the Maxwell-Voigt model, which overestimates damping. The

force displacement plots that correspond to the four models are shown in Figs. 3.9 to

3.11. It can be seen that the closest correlation with the experimental plot is achieved in

Fig. 3.11 which corresponds to the Voigt model with Bouc-Wen element and nonlinear

stiffness. This result is achieved because this particular model is the most comprehensive

model among the four models. This model contains a time varying Bouc-Wen element

and a cubic relationship to model the transition in stiffness due to snubbing. The Voigt

model with Bouc-Wen element shows similar results as the Voigt model with Bouc-Wen

element and nonlinear stiffness since it uses a time varying Bouc-Wen element as well.

But, it does not capture the transition in stiffness characteristics of the snubbing system.

The Voigt model is the easiest model among all of the four models, although it might not

be a good candidate if monitoring hysteresis and stiffness transition is the goal. It’s best

suited for application with little damping and snubbing application with low hysteresis

and limited range of excitation frequency. With regards to using the Maxwell-Voigt

model, there are no clear advantages in using it over the Voigt model. The Maxwell-

Voigt model overestimates the damping as can be seen in Table 3.1.

 43

3.5 Summary

In this chapter, mount characterization is discussed. The development of the

equations of motion for the different mounting systems to characterize the mount and

determine its parameters is presented herein. Four different models are proposed; the

Voigt model, the Maxwell-Voigt model, Voigt model with Bouc-Wen element and Voigt

model with Bouc-Wen element and nonlinear stiffness. In all of these models, mount

parameters are identified by solving an optimization problem. The objective is to

minimize the transmitted loads to the frame. The hysteresis loop for the four models is

generated and compared to the hysteresis loop generated from the experimental data. It is

seen that model 4 yields the best correlation with the experimental data.
 44

4 Chapter 4 – Mount Modeling and Design

This chapter provides the necessary mathematical equations needed to describe

the mounting system. Two different models are formulated. The first one is a six degree

of freedom model that treats the powertrain as a six DOF rigid body. The second model is

a more elaborate twelve DOF model that treats the powertrain and the swing-arm as two

6 DOF rigid bodies. The main goal is to achieve an appropriate mounting system that

fulfils the major task of vibration isolation. The models suggested above are used to

formulate the optimization problem such that the mounting system stiffness, orientation

and location are estimated. Several examples are provided based on the theoretical

models presented herein.

4.1 Mathematical Modeling

This section presents two different configurations of the engine mount system that

is used in the motorcycle vibration isolation application. The first one is a six degree of

freedom (DOF) model and the second one is a twelve DOF model. The equations of

motion are developed for both models. For both models, an optimization problem is set

up in order to solve for the engine mount characteristics by minimizing the transmitted

loads to the frame due to engine excitation loads and road loads. As mentioned above, the

optimization problem used the engine mount parameters i.e. stiffness, location and

orientation as the design vector.

4.1.1 Six DOF Model

In this section, the equations of motion for a six DOF model which captures the

engine dynamics are formulated. The model discussed in this section consists of a
 45

powertrain that is treated as a six DOF rigid body. The powertrain assembly consists of

the engine and the exhaust system connected to the frame via engine mounts. The

powertrain assembly used herein is considered to be rigid; this assumption is used

throughout this dissertation. The frame structure is also assumed to be infinitely rigid.

Fig. 4.1 shows the layout that represents the model defined above where the powertrain is

directly assembled to the frame at points (1, 2, 3) without being coupled to the swing-arm

assembly. The swing-arm is attached to the frame below point 2.

Figure 4.1: Six DOF Model (Cocco, 2001)

The equation of motion of the six DOF system defined above is given as follows:

𝑀𝑀𝑒𝑒 𝑋𝑋̈𝑒𝑒 + 𝐶𝐶𝑒𝑒 𝑋𝑋̇𝑒𝑒 + 𝐾𝐾𝑒𝑒 𝑋𝑋𝑒𝑒 = 𝐹𝐹𝑒𝑒 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑗 (4.1)

In Eq. (4.1), Fe denotes the input force vector which can be caused either by the shaking

force due to engine imbalance or due to road load. Me, Ce and Ke are 6x6 that represents
 46

mass, damping and stiffness matrices respectively. The 6x1 vector

𝑇𝑇
𝐹𝐹𝑒𝑒 = �𝑓𝑓𝑥𝑥𝑥𝑥 𝑓𝑓𝑦𝑦𝑦𝑦 𝑓𝑓𝑧𝑧𝑧𝑧 𝑚𝑚𝑥𝑥𝑥𝑥 𝑚𝑚𝑦𝑦𝑦𝑦 𝑚𝑚𝑧𝑧𝑧𝑧 � , where the first three terms in the column vector

represents the forces in the x, y, z directions and the last three terms represents the

moments about the x, y, z axes. The terms of the generalized inertia matrix Me of the

powertrain are with respect of the global coordinate system. The 6x1 displacement vector

Xe, consists of three translational x, y and z and three rotational 𝛼𝛼, 𝛽𝛽 and 𝛾𝛾 degrees of

freedom of the powertrain.

𝑋𝑋𝑒𝑒 = [𝑥𝑥 𝑦𝑦 𝑧𝑧 𝛼𝛼 𝛽𝛽 𝛾𝛾]𝑇𝑇

In order to account for different orientations of the mounts, the stiffness and

damping are represented in the local coordinate system of the mount. A transformation

matrix is used to express the stiffness and damping in the global coordinate system. The

generalized mass matrix of the powertrain is as follows (Harris, 1961):

𝑚𝑚𝑒𝑒 0 0 0 𝑚𝑚𝑒𝑒 𝑧𝑧𝑒𝑒 −𝑚𝑚𝑒𝑒 𝑦𝑦𝑒𝑒

⎡ 0 𝑚𝑚𝑒𝑒 0 −𝑚𝑚𝑒𝑒 𝑧𝑧𝑒𝑒 0 𝑚𝑚𝑒𝑒 𝑥𝑥𝑒𝑒 ⎤
⎢ ⎥
⎢ 0 0 𝑚𝑚𝑒𝑒 𝑚𝑚𝑒𝑒 𝑦𝑦𝑒𝑒 −𝑚𝑚𝑒𝑒 𝑥𝑥𝑒𝑒 0 ⎥
𝑀𝑀𝑒𝑒 = ⎢ 0 −𝑚𝑚𝑒𝑒 𝑧𝑧𝑒𝑒 𝑚𝑚𝑒𝑒 𝑦𝑦𝑒𝑒 𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 −𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 −𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 ⎥ (4.2)
⎢ 𝑚𝑚 𝑧𝑧 0 −𝑚𝑚𝑒𝑒 𝑥𝑥𝑒𝑒 −𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 𝐼𝐼𝑦𝑦𝑦𝑦𝑦𝑦 −𝐼𝐼𝑦𝑦𝑦𝑦𝑦𝑦 ⎥⎥
⎢ 𝑒𝑒 𝑒𝑒
⎣−𝑚𝑚𝑒𝑒 𝑦𝑦𝑒𝑒 𝑚𝑚𝑒𝑒 𝑥𝑥𝑒𝑒 0 −𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 −𝐼𝐼𝑦𝑦𝑦𝑦𝑦𝑦 𝐼𝐼𝑧𝑧𝑧𝑧𝑧𝑧 ⎦

In Eq. (4.2), me is the mass of the powertrain, (xe, ye, ze) is the location of the

center of gravity (C.G.) of the powertrain with respect to the origin of the global

coordinate system, Ixxe, Iyye, Izze, … are the inertia terms of the powertrain assembly with

respect to the origin of the global coordinate system.

If the center of gravity of the powertrain coincides with the origin of the global

coordinate system, the generalized mass matrix simplifies to the following form:
 47

𝑚𝑚𝑒𝑒 0 0 0 0 0
⎡ 0 𝑚𝑚𝑒𝑒 0 0 0 0 ⎤
⎢ ⎥
⎢ 0 0 𝑚𝑚𝑒𝑒 0 0 0 ⎥
𝑀𝑀𝑒𝑒 = ⎢ 0 0 0 𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 −𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 −𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 ⎥ (4.3)
⎢ 0 0 0 −𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 𝐼𝐼𝑦𝑦𝑦𝑦𝑦𝑦 −𝐼𝐼𝑦𝑦𝑦𝑦𝑦𝑦 ⎥⎥
⎢
⎣ 0 0 0 −𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 −𝐼𝐼𝑦𝑦𝑦𝑦𝑦𝑦 𝐼𝐼𝑧𝑧𝑧𝑧𝑧𝑧 ⎦

The stiffness and damping matrices 𝑘𝑘𝑖𝑖∗ and 𝑐𝑐𝑖𝑖∗ of an individual mount ‘i’ expressed in

its own local coordinate system is given as follows:

𝑘𝑘𝑥𝑥𝑥𝑥 0 0
𝑘𝑘𝑖𝑖∗ = �0 𝑘𝑘𝑦𝑦𝑦𝑦 0� (4.4)
0 0 𝑘𝑘𝑧𝑧𝑧𝑧

𝑐𝑐𝑥𝑥𝑥𝑥 0 0
𝑐𝑐𝑖𝑖∗ = �0 𝑐𝑐𝑦𝑦𝑦𝑦 0� (4.5)
0 0 𝑐𝑐𝑧𝑧𝑧𝑧

In Eq. (4.4), kxi, kyi, kzi represents the stiffness of the engine mount ‘i’ in the x, y

and z directions respectively. In the above representation it is assumed that the engine

mount is modeled about its center of elasticity which consists of three principal stiffness

coefficients without any cross coupling influence. The same can be said about the

damping matrix 𝑐𝑐𝑖𝑖∗ represented in Eq. (4.5). A transformation matrix (Ai) is used in order

to transform both, the stiffness and damping matrices to the global coordinate system as

follows:

𝑘𝑘𝑖𝑖 = 𝐴𝐴𝑇𝑇𝑖𝑖 𝑘𝑘𝑖𝑖∗ 𝐴𝐴𝑖𝑖 (4.6)

𝑐𝑐𝑖𝑖 = 𝐴𝐴𝑇𝑇𝑖𝑖 𝑐𝑐𝑖𝑖𝑇𝑇 𝐴𝐴𝑖𝑖 (4.7)

𝑐𝑐𝑖𝑖 𝑎𝑎𝑎𝑎𝑎𝑎 𝑘𝑘𝑖𝑖 in Eq. (4.6) and Eq. (4.7) respectively are the individual mount damping and

stiffness matrices expressed in the global coordinate system. The matrix Ai is a

transformation matrix which is a combination of three different rotations 𝜃𝜃1 , 𝜃𝜃2 𝑎𝑎𝑛𝑛𝑛𝑛 𝜃𝜃3

about x, y and z axes with respect to the global coordinate system shown in Eq. (4.8).
 48

Hence, for the systems where the local and global coordinate systems coincide, the

transformation matrix Ai is a 3x3 identity matrix.

𝐶𝐶𝐶𝐶2𝑖𝑖 𝐶𝐶𝐶𝐶3𝑖𝑖 −𝐶𝐶𝐶𝐶1𝑖𝑖 𝑆𝑆𝑆𝑆3𝑖𝑖 + 𝑆𝑆𝑆𝑆1𝑖𝑖 𝑆𝑆𝑆𝑆2𝑖𝑖 𝐶𝐶𝐶𝐶3𝑖𝑖 𝑆𝑆𝑆𝑆1𝑖𝑖 𝑆𝑆𝑆𝑆3𝑖𝑖 + 𝐶𝐶𝐶𝐶1𝑖𝑖 𝑆𝑆𝑆𝑆2𝑖𝑖 𝐶𝐶𝐶𝐶3𝑖𝑖
⎡ ⎤
⎢ ⎥
𝐴𝐴𝑖𝑖 = ⎢ 𝐶𝐶𝐶𝐶2𝑖𝑖 𝑆𝑆𝑆𝑆3𝑖𝑖 𝐶𝐶𝐶𝐶1𝑖𝑖 𝐶𝐶𝐶𝐶3𝑖𝑖 + 𝑆𝑆𝑆𝑆1𝑖𝑖 𝑆𝑆𝑆𝑆2𝑖𝑖 𝑆𝑆𝑆𝑆3𝑖𝑖 −𝑆𝑆𝑆𝑆1𝑖𝑖 𝐶𝐶𝐶𝐶3𝑖𝑖 + 𝐶𝐶𝐶𝐶1𝑖𝑖 𝑆𝑆𝑆𝑆2𝑖𝑖 𝑆𝑆𝑆𝑆3𝑖𝑖 ⎥ (4.8)
⎢ ⎥
⎣ −𝑆𝑆𝑆𝑆2𝑖𝑖 𝑆𝑆𝑆𝑆1𝑖𝑖 𝐶𝐶𝐶𝐶2𝑖𝑖 𝐶𝐶𝐶𝐶1𝑖𝑖 𝐶𝐶𝐶𝐶2𝑖𝑖 ⎦

where; 𝐶𝐶𝜃𝜃𝑖𝑖 = 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃𝑖𝑖 ) 𝑎𝑎𝑎𝑎𝑎𝑎 𝑆𝑆𝜃𝜃𝑖𝑖 = 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃𝑖𝑖 ). Euler angles, Bryant angles, etc. can be used

to calculate the transformation matrix Ai as well (Crede, 1965). Appendix A presents a

detailed discussion of some alternate transformation matrix formulations.

The transformed damping and stiffness matrices are for the overall six DOF

powertrain assembly is as follows:

𝐶𝐶11 𝐶𝐶12
𝐶𝐶𝑒𝑒 = � � (4.9)
𝐶𝐶21 𝐶𝐶22

𝐾𝐾11 𝐾𝐾12
𝐾𝐾𝑒𝑒 = � � (4.10)
𝐾𝐾21 𝐾𝐾22

where;

𝐾𝐾11 = ∑ 𝑘𝑘𝑖𝑖

𝐾𝐾12 = − ∑ 𝑘𝑘𝑖𝑖 𝑟𝑟̃𝑖𝑖 , 𝐾𝐾21 = 𝐾𝐾12 (4.11)

𝐾𝐾22 = − ∑ 𝑟𝑟̃𝑖𝑖 𝑘𝑘𝑖𝑖 𝑟𝑟̃𝑖𝑖

𝐶𝐶11 = ∑ 𝑐𝑐𝑖𝑖

𝐶𝐶12 = − ∑ 𝑐𝑐𝑖𝑖 𝑟𝑟̃𝑖𝑖 , 𝐶𝐶21 = 𝐶𝐶12 (4.12)

𝐶𝐶22 = − ∑ 𝑟𝑟̃𝑖𝑖 𝑐𝑐𝑖𝑖 𝑟𝑟̃𝑖𝑖

 49

𝐶𝐶𝑒𝑒 and 𝐾𝐾𝑒𝑒 represents the overall damping and stiffness matrices of the powertrain

assembly shown in Eq. (4.9) and Eq. (4.12). 𝑟𝑟̃𝑖𝑖 represents the skew-symmetric matrix that

corresponds to an individual mount position �𝑟𝑟𝑥𝑥𝑥𝑥 , 𝑟𝑟𝑦𝑦𝑦𝑦 , 𝑟𝑟𝑧𝑧𝑧𝑧 � and it is given by:

0 −𝑟𝑟𝑧𝑧𝑧𝑧 𝑟𝑟𝑦𝑦𝑦𝑦
𝑟𝑟̃𝑖𝑖 = � 𝑟𝑟𝑧𝑧𝑧𝑧 0 −𝑟𝑟𝑥𝑥𝑥𝑥 � (4.13)
−𝑟𝑟𝑦𝑦𝑦𝑦 𝑟𝑟𝑥𝑥𝑥𝑥 0

For the powertrain assembly used herein, the connection of the infinitely rigid

powertrain is done to an infinitely rigid frame through four engine mounts. The

governing equations of motion (EOM) are expanded below. It is assumed that the global

coordinate system is not located at the C.G. of the powertrain. The position vector from

the origin to the C.G. of the powertrain assembly is (xbe, ybe, zbe).

𝑚𝑚𝑒𝑒 𝑥𝑥̈ + 𝑚𝑚𝑒𝑒 𝑧𝑧𝑏𝑏𝑏𝑏 𝛽𝛽̈ + 𝑚𝑚𝑒𝑒 𝑦𝑦𝑏𝑏𝑏𝑏 𝛾𝛾̈ + (𝑘𝑘𝑥𝑥1 + 𝑘𝑘𝑥𝑥2 + 𝑘𝑘𝑥𝑥3 + 𝑘𝑘𝑥𝑥4 )𝑥𝑥 +

(𝑘𝑘𝑥𝑥1 𝑟𝑟𝑧𝑧1 + 𝑘𝑘𝑥𝑥2 𝑟𝑟𝑧𝑧2 + 𝑘𝑘𝑥𝑥3 𝑟𝑟𝑧𝑧3 + 𝑘𝑘𝑥𝑥4 𝑟𝑟𝑧𝑧4 )𝛽𝛽 −

�𝑘𝑘𝑥𝑥1 𝑟𝑟𝑦𝑦1 + 𝑘𝑘𝑥𝑥2 𝑟𝑟𝑦𝑦2 + 𝑘𝑘𝑥𝑥3 𝑟𝑟𝑦𝑦3 + 𝑘𝑘𝑥𝑥4 𝑟𝑟𝑦𝑦4 �𝛾𝛾 = 𝑓𝑓𝑥𝑥𝑥𝑥 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑗 (4.14)

𝑚𝑚𝑒𝑒 𝑦𝑦̈ − 𝑚𝑚𝑒𝑒 𝑧𝑧𝑏𝑏𝑏𝑏 𝑧𝑧̈ + 𝑚𝑚𝑒𝑒 𝑥𝑥𝑏𝑏𝑏𝑏 𝛾𝛾̈ + �𝑘𝑘𝑦𝑦1 + 𝑘𝑘𝑦𝑦2 + 𝑘𝑘𝑦𝑦3 + 𝑘𝑘𝑦𝑦4 �𝑦𝑦

− �𝑘𝑘𝑦𝑦1 𝑟𝑟𝑧𝑧1 + 𝑘𝑘𝑦𝑦2 𝑟𝑟𝑧𝑧2 + 𝑘𝑘𝑦𝑦3 𝑟𝑟𝑧𝑧3 + 𝑘𝑘𝑦𝑦4 𝑟𝑟𝑧𝑧4 �𝛼𝛼

+ �𝑘𝑘𝑦𝑦1 𝑟𝑟𝑥𝑥1 + 𝑘𝑘𝑦𝑦2 𝑟𝑟𝑥𝑥2 + 𝑘𝑘𝑦𝑦3 𝑟𝑟𝑥𝑥3 + 𝑘𝑘𝑦𝑦4 𝑟𝑟𝑥𝑥4 �𝛾𝛾 = 𝑓𝑓𝑦𝑦𝑦𝑦 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑗 (4.15)

𝑚𝑚𝑒𝑒 𝑧𝑧̈ − 𝑚𝑚𝑒𝑒 𝑦𝑦𝑏𝑏𝑏𝑏 𝛼𝛼̈ + 𝑚𝑚𝑒𝑒 𝑥𝑥𝑏𝑏𝑏𝑏 𝛽𝛽̈ + (𝑘𝑘𝑧𝑧1 + 𝑘𝑘𝑧𝑧2 + 𝑘𝑘𝑧𝑧3 + 𝑘𝑘𝑧𝑧4 )𝑧𝑧

+ �𝑘𝑘𝑧𝑧1 𝑟𝑟𝑦𝑦1 + 𝑘𝑘𝑧𝑧2 𝑟𝑟𝑦𝑦2 + 𝑘𝑘𝑧𝑧3 𝑟𝑟𝑦𝑦3 + 𝑘𝑘𝑧𝑧4 𝑟𝑟𝑦𝑦4 �𝛼𝛼

− (𝑘𝑘𝑧𝑧1 𝑟𝑟𝑥𝑥1 + 𝑘𝑘𝑧𝑧2 𝑟𝑟𝑥𝑥2 + 𝑘𝑘𝑧𝑧3 𝑟𝑟𝑥𝑥3 + 𝑘𝑘𝑧𝑧4 𝑟𝑟𝑥𝑥4 )𝛽𝛽 = 𝑓𝑓𝑧𝑧𝑧𝑧 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑗 (4.16)
 50

−𝑚𝑚𝑒𝑒 𝑧𝑧𝑏𝑏𝑏𝑏 𝑦𝑦̈ + 𝑚𝑚𝑒𝑒 𝑦𝑦𝑏𝑏𝑏𝑏 𝑧𝑧̈ + 𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 𝛼𝛼̈ − 𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 𝛽𝛽̈ − 𝐼𝐼𝑥𝑥𝑧𝑧𝑧𝑧 𝛾𝛾̈

− �𝑘𝑘𝑦𝑦1 𝑟𝑟𝑧𝑧1 + 𝑘𝑘𝑦𝑦2 𝑟𝑟𝑧𝑧2 + 𝑘𝑘𝑦𝑦3 𝑟𝑟𝑧𝑧3 + 𝑘𝑘𝑦𝑦4 𝑟𝑟𝑧𝑧4 �𝑦𝑦

+ �𝑘𝑘𝑧𝑧1 𝑟𝑟𝑦𝑦1 + 𝑘𝑘𝑧𝑧2 𝑟𝑟𝑦𝑦2 + 𝑘𝑘𝑧𝑧3 𝑟𝑟𝑦𝑦3 + 𝑘𝑘𝑧𝑧4 𝑟𝑟𝑦𝑦4 �𝑧𝑧

2 2 2 2 2 2
+ ��𝑘𝑘𝑦𝑦1 𝑟𝑟𝑧𝑧1 + 𝑘𝑘𝑧𝑧1 𝑟𝑟𝑦𝑦1 � + �𝑘𝑘𝑦𝑦2 𝑟𝑟𝑧𝑧2 + 𝑘𝑘𝑧𝑧2 𝑟𝑟𝑦𝑦2 � + �𝑘𝑘𝑦𝑦3 𝑟𝑟𝑧𝑧3 + 𝑘𝑘𝑧𝑧3 𝑟𝑟𝑦𝑦3 �

2 2
+ �𝑘𝑘𝑦𝑦4 𝑟𝑟𝑧𝑧4 + 𝑘𝑘𝑧𝑧4 𝑟𝑟𝑦𝑦4 ��𝛼𝛼

− ��𝑘𝑘𝑧𝑧1 𝑟𝑟𝑥𝑥1 𝑟𝑟𝑦𝑦1 � + �𝑘𝑘𝑧𝑧2 𝑟𝑟𝑥𝑥2 𝑟𝑟𝑦𝑦2 � + �𝑘𝑘𝑧𝑧3 𝑟𝑟𝑥𝑥3 𝑟𝑟𝑦𝑦3 � + �𝑘𝑘𝑧𝑧4 𝑟𝑟𝑥𝑥4 𝑟𝑟𝑦𝑦4 ��𝛽𝛽

− ��𝑘𝑘𝑦𝑦1 𝑟𝑟𝑥𝑥1 𝑟𝑟𝑧𝑧1 � + �𝑘𝑘𝑦𝑦2 𝑟𝑟𝑥𝑥2 𝑟𝑟𝑧𝑧2 � + �𝑘𝑘𝑦𝑦3 𝑟𝑟𝑥𝑥3 𝑟𝑟𝑧𝑧3 � + �𝑘𝑘𝑦𝑦4 𝑟𝑟𝑥𝑥4 𝑟𝑟𝑧𝑧4 ��𝛾𝛾

= 𝑚𝑚𝑥𝑥𝑥𝑥 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑗 (4.17)

𝑚𝑚𝑒𝑒 𝑧𝑧𝑏𝑏𝑏𝑏 𝑥𝑥̈ − 𝑚𝑚𝑒𝑒 𝑥𝑥𝑏𝑏𝑏𝑏 𝑧𝑧̈ − 𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 𝛼𝛼̈ + 𝐼𝐼𝑦𝑦𝑦𝑦𝑦𝑦 𝛽𝛽̈ − 𝐼𝐼𝑦𝑦𝑦𝑦𝑦𝑦 𝛾𝛾̈

+ (𝑘𝑘𝑥𝑥1 𝑟𝑟𝑧𝑧1 + 𝑘𝑘𝑥𝑥2 𝑟𝑟𝑧𝑧2 + 𝑘𝑘𝑥𝑥3 𝑟𝑟𝑧𝑧3 + 𝑘𝑘𝑥𝑥4 𝑟𝑟𝑧𝑧4 )𝑥𝑥

− (𝑘𝑘𝑧𝑧1 𝑟𝑟𝑥𝑥1 + 𝑘𝑘𝑧𝑧2 𝑟𝑟𝑥𝑥2 + 𝑘𝑘𝑧𝑧3 𝑟𝑟𝑥𝑥3 + 𝑘𝑘𝑧𝑧4 𝑟𝑟𝑥𝑥4 )𝑧𝑧

− ��𝑘𝑘𝑧𝑧1 𝑟𝑟𝑦𝑦1 𝑟𝑟𝑥𝑥1 � + �𝑘𝑘𝑧𝑧2 𝑟𝑟𝑦𝑦2 𝑟𝑟𝑥𝑥2 � + �𝑘𝑘𝑧𝑧3 𝑟𝑟𝑦𝑦3 𝑟𝑟𝑥𝑥3 � + �𝑘𝑘𝑧𝑧4 𝑟𝑟𝑦𝑦4 𝑟𝑟𝑥𝑥4 ��𝛼𝛼

2 2 ) 2 2 ) 2 2 )
+ [(𝑘𝑘𝑥𝑥1 𝑟𝑟𝑧𝑧1 + 𝑘𝑘𝑧𝑧1 𝑟𝑟𝑥𝑥1 + (𝑘𝑘𝑥𝑥2 𝑟𝑟𝑧𝑧2 + 𝑘𝑘𝑧𝑧2 𝑟𝑟𝑥𝑥2 + (𝑘𝑘𝑥𝑥3 𝑟𝑟𝑧𝑧3 + 𝑘𝑘𝑧𝑧3 𝑟𝑟𝑥𝑥3
2 2 )]𝛽𝛽
+ (𝑘𝑘𝑥𝑥4 𝑟𝑟𝑧𝑧4 + 𝑘𝑘𝑧𝑧4 𝑟𝑟𝑥𝑥4

− ��𝑘𝑘𝑥𝑥1 𝑟𝑟𝑦𝑦1 𝑟𝑟𝑧𝑧1 � + �𝑘𝑘𝑥𝑥2 𝑟𝑟𝑦𝑦2 𝑟𝑟𝑧𝑧2 � + �𝑘𝑘𝑥𝑥3 𝑟𝑟𝑦𝑦3 𝑟𝑟𝑧𝑧3 � + �𝑘𝑘𝑥𝑥4 𝑟𝑟𝑦𝑦4 𝑟𝑟𝑧𝑧4 ��𝛾𝛾

= 𝑚𝑚𝑦𝑦𝑦𝑦 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑗 (4.18)

 51

−𝑚𝑚𝑒𝑒 𝑦𝑦𝑏𝑏𝑏𝑏 𝑥𝑥̈ + 𝑚𝑚𝑒𝑒 𝑥𝑥𝑏𝑏𝑏𝑏 𝑦𝑦̈ – 𝐼𝐼𝑥𝑥𝑥𝑥𝑥𝑥 𝛼𝛼̈ − 𝐼𝐼𝑦𝑦𝑦𝑦𝑦𝑦 𝛽𝛽̈ + 𝐼𝐼𝑧𝑧𝑧𝑧𝑧𝑧 𝛾𝛾̈

− �𝑘𝑘𝑥𝑥1 𝑟𝑟𝑦𝑦1 + 𝑘𝑘𝑥𝑥2 𝑟𝑟𝑦𝑦2 + 𝑘𝑘𝑥𝑥3 𝑟𝑟𝑦𝑦3 + 𝑘𝑘𝑥𝑥4 𝑟𝑟𝑦𝑦4 �𝑥𝑥

+ �𝑘𝑘𝑦𝑦1 𝑟𝑟𝑥𝑥1 + 𝑘𝑘𝑦𝑦2 𝑟𝑟𝑥𝑥2 + 𝑘𝑘𝑦𝑦3 𝑟𝑟𝑥𝑥3 + 𝑘𝑘𝑦𝑦4 𝑟𝑟𝑥𝑥4 �𝑦𝑦

− ��𝑘𝑘𝑦𝑦1 𝑟𝑟𝑥𝑥1 𝑟𝑟𝑧𝑧1 � + �𝑘𝑘𝑦𝑦2 𝑟𝑟𝑥𝑥2 𝑟𝑟𝑧𝑧2 � + �𝑘𝑘𝑦𝑦3 𝑟𝑟𝑥𝑥3 𝑟𝑟𝑧𝑧3 � + �𝑘𝑘𝑦𝑦4 𝑟𝑟𝑥𝑥4 𝑟𝑟𝑧𝑧4 ��𝛼𝛼

− ��𝑘𝑘𝑥𝑥1 𝑟𝑟𝑦𝑦1 𝑟𝑟𝑧𝑧1 � + �𝑘𝑘𝑥𝑥2 𝑟𝑟𝑦𝑦2 𝑟𝑟𝑧𝑧2 � + �𝑘𝑘𝑥𝑥3 𝑟𝑟𝑦𝑦3 𝑟𝑟𝑧𝑧3 � + �𝑘𝑘𝑥𝑥4 𝑟𝑟𝑦𝑦4 𝑟𝑟𝑧𝑧4 ��𝛽𝛽

2 2 2 2 2 2
+ ��𝑘𝑘𝑥𝑥1 𝑟𝑟𝑦𝑦1 + 𝑘𝑘𝑦𝑦1 𝑟𝑟𝑥𝑥1 � + �𝑘𝑘𝑥𝑥2 𝑟𝑟𝑦𝑦2 + 𝑘𝑘𝑦𝑦2 𝑟𝑟𝑥𝑥2 � + �𝑘𝑘𝑥𝑥3 𝑟𝑟𝑦𝑦3 + 𝑘𝑘𝑦𝑦3 𝑟𝑟𝑥𝑥3 �

2 2
+ �𝑘𝑘𝑥𝑥4 𝑟𝑟𝑦𝑦4 + 𝑘𝑘𝑦𝑦4 𝑟𝑟𝑥𝑥4 �� 𝛾𝛾 = 𝑚𝑚𝑧𝑧𝑧𝑧 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑗 (4.19)

In the equations presented above, Eq. (4.14) to Eq. (4.19), the damping

coefficients has been ignored in order to simplify the equations of motion. It may noted

that the transformation matrices A1, A2, A3 and A4 are represented as identity matrices

since the local frame of the individual mounts are aligned with the global coordinate

system.

The equations of motion presented above are for a motorcycle application in

which the powertrain is directly mounted to the frame as shown in Fig. 4.1. The swing-

arm assembly used in this model is not connected to the powertrain. Fig. 4.1 shows the

connection points that connect the powertrain to the frame. This type of connection is

widely used in the motorcycle industry.

4.1.2 Twelve DOF Model

A six DOF model is used to represent the powertrain assembly that is attached to

the frame through engine mounts has been discussed in the previous section. The model

developed in section 4.1.1 could come out short in capturing the isolation characteristics
 52

in this layout. Therefore, an alternative model shown in Fig. 4.2, will be developed in this

section. This model is a twelve DOF system that takes into consideration a layout that is

widely used in the motorcycle industry. This layout assumes that there are two rigid

bodies, one represents the powertrain and the second one represents the swing-arm. The

swing arm is pivoted to the powertrain through a shaft assembly referred to as the

coupler. Fig. 4.3 shows the layout of the twelve DOF model with the two rigid bodies

attached. In this section, the EOM of the 12 DOF model are developed. More details

regarding the coupler shaft are presented in Appendix C.

The general equations of motion for the twelve DOF system described above are

as follows:

𝑀𝑀 𝑋𝑋̈ + 𝐶𝐶 𝑋𝑋̇ + 𝐾𝐾 𝑋𝑋 = 𝐹𝐹 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑗 (4.20)

In Eq. (4.20), M, C and K are a 12x12 mass, damping and stiffness matrices respectively.

𝑋𝑋 = [𝑥𝑥𝑠𝑠𝑠𝑠 𝑦𝑦𝑠𝑠𝑠𝑠 𝑧𝑧𝑠𝑠𝑠𝑠 𝛼𝛼𝑠𝑠𝑠𝑠 𝛽𝛽𝑠𝑠𝑠𝑠 𝛾𝛾𝑠𝑠𝑠𝑠 𝑥𝑥𝑒𝑒 𝑦𝑦𝑒𝑒 𝑧𝑧𝑒𝑒 𝛼𝛼𝑒𝑒 𝛽𝛽𝑒𝑒 𝛾𝛾𝑒𝑒 ]𝑇𝑇 is the displacement vector

that contains translational and rotational degrees of freedom for both the swing-arm and

the powertrain. The subscript ‘sa’ represents parameters related to the swing-arm

assembly and the subscript ‘e’ represents parameters related to the powertrain assembly.

F denotes the input force vector due to the shaking force resulting from engine imbalance

and/or the road loads due to the irregularities in the road profile. The overall mass matrix

of the system is as follows:

𝑀𝑀𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑍𝑍6
𝑀𝑀 = � � (4.21)
𝑍𝑍6 𝑀𝑀𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒

In Eq. (4.21), Mengine, Mswingarm are the 6x6 mass matrices of the powertrain and the swing-

arm assemblies respectively and Z6 is a 6x6 zero matrix. The powertrain mass matrix and
 53

the swing-arm mass matrix are similar and they are the same as the mass matrix defined

in Eq. (4.20). The inertia matrices for both the powertrain and the swing-arm are defined

at their local center of gravity.

The stiffness and damping matrices of the twelve DOF system are defined as

follows:

𝐾𝐾𝑠𝑠𝑠𝑠 + 𝐾𝐾𝑐𝑐 −𝐾𝐾𝑐𝑐

𝐾𝐾 = � � (4.22)
−𝐾𝐾𝑐𝑐 𝐾𝐾𝑒𝑒 + 𝐾𝐾𝑐𝑐

𝐶𝐶𝑠𝑠𝑠𝑠 + 𝐶𝐶𝑐𝑐 −𝐶𝐶𝑐𝑐

𝐶𝐶 = � � (4.23)
−𝐶𝐶𝑐𝑐 𝐶𝐶𝑒𝑒 + 𝐶𝐶𝑐𝑐

In Eq. (4.22), Ke and Kse are the stiffness matrices of the powertrain and the swing-arm

respectively. Both of these stiffness matrices are constructed in the same fashion as

described in the previous section. The swing-arm stiffness matrix is constructed using the

stiffness characteristics of the two rear shock springs connecting the swing-arm to the

frame. Ce and Csa shown in Eq. (4.23) are the damping matrices of the powertrain and the

swing-arm respectively and they are constructed in the same way as Ke and Ksa. Kc and Cc

are the stiffness and damping matrices of the coupler respectively. Both of these matrices

are 6x6 diagonal matrices. The construction of coupling stiffness and damping matrices

of the coupler is discussed in details in Appendix C. A modeling of the shaking forces

and the road loads will be discussed in detail in chapter 5.

The model discussed in this section assumes that the frame is infinitely rigid just

like the assumption in the previous section. This assumption means that connection

points between the engine mounts and the frame as well as the connection points between

the frame and the rear suspension undergo zero deflection.

 54

Figure 4.2: Twelve DOF Model (Cocco, 2001)

 55

Mount

System

Engine Assembly

Shock 𝑦𝑦𝑒𝑒
𝑥𝑥𝑒𝑒
Assembly
𝑦𝑦𝑔𝑔

𝑧𝑧𝑒𝑒
𝑥𝑥𝑔𝑔 Mount

System

𝑦𝑦𝑠𝑠𝑠𝑠 𝑧𝑧𝑔𝑔
𝑥𝑥𝑠𝑠𝑠𝑠
Swing-arm

Assembly
𝑧𝑧𝑠𝑠𝑠𝑠

Tire Patch

Figure 4.3: Twelve DOF Engine-Swingarm Layout

 56

4.2 Mount Optimization

In this section, a formulation of the optimization problem for the models discussed

in section 4.1 is presented using. The optimization problem is solved using the method of

Sequential Quadratic Programming (SQP). SQP is used throughout this dissertation to

solve the engine mount optimization problem.

The objective function that is used in this work is the weighted sum of the

transmitted force through each individual mount. The transmitted forces through the

mounts are due to the shaking forces generated inside the engine and/or the forces

generated from the varying road profile. Loads calculated at several steady speeds can be

used to construct the objective function.

The force ‘fi’ transmitted to the frame through the individual mount ‘i’ is given as

follows:

𝑋𝑋𝑡𝑡𝑡𝑡
𝑓𝑓𝑖𝑖 = [−𝑘𝑘𝑖𝑖∗ 𝑘𝑘𝑖𝑖∗ 𝑟𝑟̃𝑖𝑖 ] � � (4.24)
𝑋𝑋𝑟𝑟𝑟𝑟

In Eq. (4.24), 𝑋𝑋𝑡𝑡𝑡𝑡 and 𝑋𝑋𝑟𝑟𝑟𝑟 represent the translational and rotational displacement at the

center of gravity of the powertrain as result of the input load. 𝑘𝑘𝑖𝑖∗ is the local stiffness

matrix for the individual mount ‘i’ and 𝑟𝑟̃𝑖𝑖 is the skew symmetric matrix from the position

vector of the individual mount ‘i’. Both of these matrices are defined in section 4.1.1 by

Eq. (4.4) and Eq. (4.13).

The objective function 𝑓𝑓𝑤𝑤 is assembled by summing the Euclidean norm of the

individual force transmitted through each mount as follows:

𝑓𝑓𝑤𝑤 = � 𝜆𝜆𝑗𝑗 �‖𝑓𝑓𝑖𝑖 ‖ (4.25)

𝑗𝑗 𝑖𝑖
 57

In Eq. (4.25), 𝜆𝜆𝑖𝑖 is the weighting parameter that corresponds to different loading

conditions. The complete engine mount optimization problem can be stated as follows:

𝑀𝑀𝑀𝑀𝑀𝑀 𝑓𝑓𝑤𝑤 (𝑘𝑘𝑖𝑖 , 𝑟𝑟𝑖𝑖 , 𝜃𝜃𝑖𝑖 )

𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑡𝑡𝑡𝑡 𝑔𝑔𝑗𝑗 (𝑘𝑘𝑖𝑖 , 𝑟𝑟𝑖𝑖 , 𝜃𝜃𝑖𝑖 ) ≤ 0 𝑗𝑗 = 1, … , 𝑁𝑁 (4.26)

In Eq. (4.26), the mount stiffness, location and orientation (𝑘𝑘𝑖𝑖 , 𝑟𝑟𝑖𝑖 , 𝜃𝜃𝑖𝑖 ) are the design

variables that are subjected to a total of N number of constraints 𝑔𝑔𝑗𝑗 . The constraints that

are used in the above problem consist of constraints on the engine mount stiffness,

constraints on the mount location based on the available space, constraints on the mount

orientation that is dictated by symmetry and finally a constraint on the deflection of the

center of gravity of the powertrain due to the static weight of the powertrain. The

objective function fw is defined in Eq. (4.25). Both fw and 𝑔𝑔𝑗𝑗 are functions of the design

variables (𝑘𝑘𝑖𝑖 , 𝑟𝑟𝑖𝑖 , 𝜃𝜃𝑖𝑖 ).

The Sequential Quadratic Programming (SQP) method is used to solve the

optimization problem. A brief description of the SQP method is presented next.

4.2.1 Sequential Quadratic Programming

The general function of an optimization problem is given below:

𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 𝑓𝑓(𝑥𝑥); 𝑥𝑥 ∈ ℛ 𝑛𝑛

𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑡𝑡𝑡𝑡 ℎ𝑖𝑖 (𝑥𝑥) = 0 𝑖𝑖 = 1, … , 𝑝𝑝 (4.27)

𝑔𝑔𝑗𝑗 (𝑥𝑥) ≤ 0 𝑗𝑗 = 1, … , 𝑚𝑚

𝑥𝑥𝑙𝑙 ≤ 𝑥𝑥 ≤ 𝑥𝑥𝑢𝑢

In Eq. (4.27), f(x) is the objective function. hi(x) and gi(x) are the ith equality and

inequality constraints respectively and x is the vector of design variables. xl and xu are the

lower and upper bound vectors for the design variables. The optimization problem
 58

defined in Eq. (4.27) consisting of ‘p’ equality constraints and ‘m’ inequality constraints

is said to be linear if the objective function and all the constraints are linear function of

the design variables. The problem is said to be quadratic if the objective function is

quadratic and the constraints are linear. If the objective function and/or the constraints are

nonlinear function of the design variables, then the problem is said to be a nonlinear

optimization problem. In the case of the engine mount optimization, the problem is

nonlinear, and the Sequential Quadratic Programming (SQP) algorithm is employed to

solve the problem. The (SQP) method uses the Newton’s method and Kuhn-Tucker

conditions to solve the optimization problem as mentioned in (Rao, 2000).

For the optimization problem with p equality constraints and n design variables, a

quadratic sub-problem is constructed based on the approximation of the Lagrangian

function L(x, λ) which stated as follows:

𝑝𝑝

𝐿𝐿(𝑥𝑥, 𝜆𝜆) = 𝑓𝑓(𝑥𝑥) + � 𝜆𝜆𝑘𝑘 ℎ𝑘𝑘 (𝑥𝑥) (4.28)

𝑘𝑘=1

In Eq. (4.28), λk is the Lagrange multiplier for the kth equality constraint and hk(x) is the

kth equality constraint and f is the objective function. x is a vector of n design variables.

The Kuhn-Tucker necessary conditions for the problem stated in Eq. (4.28) can be

stated as follows:
𝑝𝑝

∇𝐿𝐿 = 0 or ∇𝑓𝑓 + � 𝜆𝜆𝑘𝑘 ∇ℎ𝑘𝑘 = 0

𝑘𝑘=1

ℎ𝑘𝑘 (𝑥𝑥) = 0, 𝑘𝑘 = 1, … , 𝑝𝑝 (4.29)

Eq. (4.29) consists of a set of (n + p) nonlinear equations that is solved using Newton’s

method. The above equation can be represented in the following form:

 59

∇𝐿𝐿
𝐹𝐹(𝑌𝑌) = 0 where 𝐹𝐹 = � � and
ℎ (𝑛𝑛+𝑝𝑝)𝑥𝑥1

𝑥𝑥
𝑌𝑌 = � � (4.30)
𝜆𝜆 (𝑛𝑛 +𝑝𝑝)𝑥𝑥1

where the (n + p) system of equation shown in Eq. (4.30) is solved using the Newton’s

method iteratively as follows:

[∇𝐹𝐹]𝑇𝑇𝑗𝑗 ∆𝑌𝑌𝑗𝑗 = −𝐹𝐹�𝑌𝑌𝑗𝑗 � (4.31)

In Eq. (4.31) Yj is the solution at the beginning of the the jth iteration, ΔYj is the change in

the Yj and [𝛻𝛻𝐹𝐹]𝑗𝑗 is the Jacobian matrix of the (n + p) nonlinear equations. The updated

solution is given as follows:

𝑌𝑌𝑗𝑗 +1 = 𝑌𝑌𝑗𝑗 + ∆𝑌𝑌𝑗𝑗 (4.32)

Eq. (4.32) can be rewritten as:

[∇2 𝐿𝐿] [𝐻𝐻] ∆𝑥𝑥 ∇𝐿𝐿

� � � � = −� � (4.33)
[𝐻𝐻]𝑇𝑇 [0] 𝑗𝑗 ∆𝜆𝜆 𝑗𝑗 ℎ 𝑗𝑗

In Eq. (4.33), ∆𝑥𝑥𝑗𝑗 = 𝑥𝑥𝑗𝑗 +1 − 𝑥𝑥𝑗𝑗 and ∆𝜆𝜆𝑗𝑗 = 𝜆𝜆𝑗𝑗 +1 − 𝜆𝜆𝑗𝑗 and ∇2 𝐿𝐿 is the Hessian matrix of

the Lagrange function and 𝐻𝐻 = [∇ℎ𝑘𝑘 ]. The first equation from the system of equation in

Eq. (4.33) can be rewritten in the following form:

[∇2 𝐿𝐿]𝑗𝑗 Δ𝑥𝑥𝑗𝑗 + [𝐻𝐻]𝑗𝑗 𝜆𝜆𝑗𝑗 +1 = −∇𝐿𝐿𝑗𝑗 + [𝐻𝐻]𝑗𝑗 𝜆𝜆𝑗𝑗 = −∇𝑓𝑓𝑗𝑗 (4.34)

Eq. (4.33) can be rewritten as:

[∇2 𝐿𝐿] [𝐻𝐻] ∆𝑥𝑥𝑗𝑗 ∇𝑓𝑓𝑗𝑗

� � � � = −� � (4.35)
[𝐻𝐻]𝑇𝑇 [0] 𝑗𝑗
𝜆𝜆𝑗𝑗 +1 ℎ𝑗𝑗
 60

Eq. (4.35) can be solved iteratively to determine ∆𝑥𝑥𝑗𝑗 and 𝜆𝜆𝑗𝑗 +1 , the design variables and

the Lagrange multipliers. For a general problem with both equality and inequality

constraints, the optimization problem can be stated as follows:

Find 𝑋𝑋 that

1
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 𝑄𝑄 = ∇𝑓𝑓 𝑇𝑇 ∆𝑋𝑋 + ∆𝑥𝑥 𝑇𝑇 [∇2 𝐿𝐿]∆𝑋𝑋
2

𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑡𝑡𝑡𝑡 𝑔𝑔𝑗𝑗 + ∇𝑔𝑔𝑗𝑗𝑇𝑇 ∆𝑋𝑋 ≤ 0 𝑗𝑗 = 1, … , 𝑚𝑚 (4.36)

ℎ𝑘𝑘 + ∇ℎ𝑘𝑘𝑇𝑇 ∆𝑋𝑋 = 0 𝑘𝑘 = 1, … , 𝑝𝑝

and the Lagrangian function is given as:

𝑚𝑚 𝑝𝑝

𝐿𝐿� = 𝑓𝑓(𝑥𝑥) + � 𝜆𝜆𝑗𝑗 𝑔𝑔𝑗𝑗 (𝑥𝑥) + � 𝜆𝜆𝑚𝑚 +𝑘𝑘 ℎ𝑘𝑘 (𝑥𝑥) (4.37)
𝑗𝑗 =1 𝑘𝑘=1

In Eq. (4.37), X is the design variable vector and L is the corresponding Lagrangian

function. 𝑔𝑔𝑗𝑗 and ℎ𝑘𝑘 are the inequality and equality constraints respectively. A first order

Taylor series is used to linearize the nonlinear constraint function. This problem can be

solved with a similar procedure as the optimization problem with only equality

constraints mentioned earlier in this section.

4.2.2 Six DOF Model

In this section, the force transmitted through the engine mount to the frame due to

the engine imbalance is used as the objective function. The optimization problem is

formulated based on the six DOF model presented in section 4.1.1. The design vector is

based on the mount parameters; stiffness, orientation and location. Some constraints

imposed on the problem include limits on the powertrain deflection due to the static and
 61

dynamic loads combined with constraints on the lower and upper bounds for the design

variables.

4.2.2.1 Numerical Example

The example presented herein is based on the model presented in section 4.1.1 in

order to solve the optimization problem of minimizing the transmitted loads formulated

in section 4.2. The objective function is compiled by summing the transmitted force

through the individual mounts that are supporting the powertrain. The objective function

is described as follows:

𝐹𝐹𝑜𝑜𝑜𝑜𝑜𝑜 = ‖𝑓𝑓1 ‖ + ‖𝑓𝑓2 ‖ + ‖𝑓𝑓3 ‖ + ‖𝑓𝑓4 ‖ (4.38)

In Eq. (4.38), f1, f2, f3 and f4 are the force vectors transmitted to the frame through the four

mounts that are supporting the powertrain due to the shaking force at the engine steady

speed of 4000 rpm. The formulation of the shaking force at the steady state speed is

discussed in depth in chapter 5. The general layout of the mounting system for this

example is shown in Fig. 4.4. Mount parameters which consist of mount stiffness, mount

location and mount orientation are compiled to form the design vector. The lower and

upper bounds used for the design variables are shown in Table 4.1. A limit that is

imposed on the design variables by constraining the deflection of the powertrain as

follows:

|𝑈𝑈𝑠𝑠𝑠𝑠 | ≤ 𝑈𝑈𝑚𝑚𝑚𝑚𝑚𝑚 (4.39)

In Eq. (4.39), Ust is the static deflection vector of the powertrain due to the static loading

at its C.G. and Umax is the maximum allowable displacement due to the static load.

𝑈𝑈𝑚𝑚𝑚𝑚𝑚𝑚 = [0.025 𝑖𝑖𝑖𝑖 0.050 in 0.025 in 0.5 𝑑𝑑𝑑𝑑𝑑𝑑 0.5 𝑑𝑑𝑑𝑑𝑑𝑑 0.5 𝑑𝑑𝑑𝑑𝑑𝑑]. An

additional constraint is added to the allowable displacement at the mount in the y-

 62

direction to prevent premature snubbing. A maximum displacement of 0.5 in is used as an

upper bound for the displacement for all four mount locations. The maximum steady state

displacement is as follows:

𝑑𝑑𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑𝑑𝑑

|𝑈𝑈1 |𝑑𝑑𝑑𝑑𝑑𝑑 ≤ 𝑈𝑈𝑚𝑚𝑚𝑚𝑚𝑚 |𝑈𝑈2 |𝑑𝑑𝑑𝑑𝑑𝑑 ≤ 𝑈𝑈𝑚𝑚𝑚𝑚𝑚𝑚 |𝑈𝑈3 |𝑑𝑑𝑑𝑑𝑑𝑑 ≤ 𝑈𝑈𝑚𝑚𝑚𝑚𝑚𝑚 |𝑈𝑈4 |𝑑𝑑𝑑𝑑𝑑𝑑 ≤ 𝑈𝑈𝑚𝑚𝑚𝑚𝑚𝑚

In this example, the mounting system used herein consists of four identical

circular cross section elastomeric mounts. For each mount, two dynamic stiffness

parameters completely define the stiffness characteristics. These stiffness parameters are

radial and axial stiffness and are used as the design variables. A loss factor of 0.3

(Carfagni, 1998) and a dynamic-to-static stiffness coefficient of 1.2 have been used. In

order to reduce the total number of design variables, symmetry constraints are imposed.

This is done by symmetrically placing two mounts on each side of the x-y plane resulting

in six position variables instead of twelve and four orientation variables instead of twelve.

The radial and axial stiffness values are identical for all four mounts resulting in a total of

twelve design variable for the engine mounting system. The mass of the powertrain is 0.5

lb-s2/in and inertia values of the powertrain are given in Table 4.2.

The optimization problem is solved using the SQP technique to minimize the

value of the objective function. The design variables resulting from the optimization

process are shown in Table 4.3. The computed mount location and orientation vectors are

shown in Table 4.4. The resulting mode shapes are shown in Fig. 4.5. Each mode shape is

presented with its corresponding un-damped natural frequency with respect to their

degrees of freedom 𝑥𝑥, 𝑦𝑦, 𝑧𝑧, 𝛼𝛼, 𝛽𝛽 and 𝛾𝛾 respectively.

 63

Top View

Rear End C.G x Front End

of motorcycle of motorcycle
z

y
x

C.G

Front View

�𝑥𝑥2 , 𝑦𝑦2 , 𝑧𝑧2 �

�𝑥𝑥1 , 𝑦𝑦1 , 𝑧𝑧1 �

Figure 4.4: Mount System Layout

 64

1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 10.8978Hz Mode corresponding to 16.0781Hz
1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 20.8461Hz Mode corresponding to 28.8037Hz
1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 48.741Hz Mode corresponding to 58.1264Hz

Figure 4.5: Mode Shapes for the Optimized Configuration (6 DOF Model)

Table 4.1: Bounds for Design Variables

Min. Max.
Mount Stiffness
100 5000
(x,y) lb/in
Mount Stiffness (z) 500 10000

Orientation Angles deg. 0 50

Table 4.2: Inertia Tensor of Powertrain Assembly

x y Z

Ix 20.7 1.86 0.12

Iy (lb-in2) 1.86 12.81 2.3

Iz 0.12 2.3 26.14

 65

Table 4.3: Optimization Results (6 DOF Model)

Load
Mount Stiffness (lb/in)
Transmitted
(lb) x y z

Initial Guess 110.90 475 475 7500

Optimized Design 39.90 1016.6 1016.6 7503.5

58.126 (un-
Natural Frequencies 10.898 16.078 20.846 28.804 48.741
damped)
57.627
(Hz) 10.89 16.07 20.823 28.743 48.447
(damped)
Damping
0.025 0.036 0.047 0.065 0.110 0.131
Coefficients

Table 4.4: Optimization Results for Position and Orientation (6 DOF Model)

Mount 1 Mount 2 Mount 3 Mount 4

Orientation (deg)
(0.1, 50, 0) (-0.1, -50, 0) (0.5, 25, 0) (-0.5, -25, 0)
Starting Guess
Results (16.6, 50, 0) (-16.4, -50, 0) (45.7, 12.7, 0) (-45.7, -12.7, 0)

Position (in)
(12, -9, 0) (12, -9, 0) (-19, -5, 0) (-19, -5, 0)
Starting Guess
Results (9.2, -5, -7) (9.2, -5, 7) (-11, -10, -3) (-11, -10, 3)

4.2.2.2 Discussion of Results

It is worth mentioning that for the example presented in section 4.2.2.1, the

starting guess for the problem was changed couple of times to make sure that the final

optimum solution does not get stuck at local minima. This is due to the fact that the

engine mount optimization problem is a highly nonlinear and could easily get stuck at a

local minimum as a final solution.

 66

As can be seen in Table 4.3, the optimum stiffness values have changed a lot in

the x and y direction while barely changing in the z-direction. The small amount of

change in the stiffness along the z-direction is due to the fact that the force transmitted to

the frame is least sensitive to the stiffness in the z-direction of the engine mount. This

fact is used in the motorcycle design by tuning the (out-of-plane) stiffness in the z-

direction to achieve the best handling possible. By tuning the (in-plane) stiffness values

in the x and y direction, the isolation characteristics of the motorcycle are enhanced with

minimal cross coupling between the out-of-plane and the in-plane stiffness coefficients.

Table 4.4, shows the optimum values for the mount locations and orientations. It can be

seen that the mount locations have not changed a lot, meanwhile the mount orientations

have changed significantly. This observation shows the effect of mount orientation on

achieving minimum load transmission. On the other hand it also indicates that the effect

of mount location on load transmission has less impact than mount orientation.

4.2.3 Twelve DOF Model

The equations of motion of the six DOF model formulated in section 4.1.1 along

with the example shown in section 4.1.2 represents a mounting system that is connected

to the frame only. Fig. 4.2 shows an alternate twelve DOF model that couples the

powertrain and the swing-arm using a shaft assembly. The model presented herein is an

extension of the model presented in section 4.2.2. It provides sufficient information to

capture the isolation characteristics of such a layout. This model is based on two rigid

bodies, one is for the powertrain assembly and the other one is for the swing-arm

assembly connected together using a coupler.

 67

This section uses the equations that were formulated in section 4.1.2 to develop an

optimization problem in order to solve for the mount parameters. It will also provide

examples where the road loads and the shaking loads are present as input loads.

4.2.3.1 Numerical Example I

The example presented next is based on the model presented in section 4.1.2 in

order to solve the optimization problem of minimizing the transmitted loads formulated

in section 4.2. The objective function is computed by summing the transmitted force

through the individual mounts that are supporting the powertrain. The objective function

is described in Eq. (4.38). The input load vector corresponds to the force due to the

shaking loads only.

The mounting system used in this example consists of four identical circular cross

section elastomeric mounts with symmetry constraints. Two of these engine mount are at

the front of the powertrain assembly and the other two are located at the rear of the

powertrain assembly as shown in Fig. 4.4. The powertrain assembly and the swing-arm

assembly are connected using a shaft assembly which will be referred to as the coupler.

The swing-arm assembly is connected to the frame via two shock absorbers one at each

side of the motorcycle.

The swing arm assembly used herein (Kaul, 2006) has a mass of 0.13 lb-s2/in. the

inertia properties of the swing arm with respect to its C.G. are listed in Table 4.5. The

swing-arm is connected to the frame using two shock absorbers which are inclined by an

angle of 47o with respect to the horizontal axis. The shock absorber exhibits an axial

stiffness and damping of 45 lb/in and 4.4 lb-s/in respectively. The stiffness of the coupler

used in the example is 42655 lb/in in the x and y direction and 658252 lb/in along the z
 68

axis. The rotational stiffness values is 682493 lb-in/rad about the x and y axes. The

rotational stiffness about the z axis is zero. A 2% structural damping has been used to

compute the coupler damping properties. The input load, the design parameters, bounds

and constraints are the same as the example presented in section 4.2.2.1. The results of

the optimization problem are presented in Table 4.6 and Table 4.7.

Table 4.5: Inertia Tensor of the Swing-arm Assembly

x y z

Ix 0.465 0.002 -0.007

Iy (lb-in2) 0.002 30 -0.008

Iz -0.007 -0.008 29

Table 4.6: Optimization Results (12 DOF Model – Shaking Load only)

Load Transmitted Mount Stiffness (lb/in)

(lb) x y Z

Initial Guess 88.59 475 475 7500

Optimized
59.32 2341.7 2341.7 2157.9
Design
1.640 8.719 10.436 21.394 22.221 24.305
Natural
Frequencies 1271.369
59.842 75.703 103.415 104.025 200.567
(undamped)
1.574 8.618 10.426 21.375 22.217 24.244
(Hz) 1270.350
59.3309 75.119 103.373 104.024 198.376
(damped)

Damping 0.005 0.020 0.028 0.040 0.042 0.044

Coefficients 0.071 0.124 0.130 0.147 0.152 0.282
 69

Table 4.7: Optimization Results for Position and Orientation (12 DOF Model – Shaking
Load only)

Mount 1 Mount 2 Mount 3 Mount 4

Orientation (deg)
(0.1, 50, 0) (-0.1, -50, 0) (0.5, 25, 0) (-0.5, -25, 0)
Starting Guess
Results (50, 50, 0) (-50, -50, 0) (50, 50, 0) (-50, -50, 0)
Position (in)
(12, -9, 0) (12, -9, 0) (-19, -5, 0) (-19, -5, 0)
Starting Guess
Results (12, -9,-3) (12, -9,3) (-12.5, -10, -7) (-12.5, -10, 7)

1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 1.5102Hz Mode corresponding to 10.7701Hz
1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 11.996Hz Mode corresponding to 19.7639Hz
1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 21.9069Hz Mode corresponding to 24.6917Hz

Figure 4.6: Mode Shapes - 1 to 6 (12 DOF Model – Shaking Load only)
 70

1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 72.565Hz Mode corresponding to 103.1124Hz
1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 104.6553Hz Mode corresponding to 201.053Hz
1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 207.5377Hz Mode corresponding to 1271.3505Hz

Figure 4.7: Mode Shapes - 7 to 12 (12 DOF Model – Shaking Load only)

4.2.3.2 Numerical Example II

This example is based on the model presented in section 4.1.2 in order to solve

the optimization problem of minimizing the transmitted loads formulated in section 4.2.

The objective function is computed by summing the transmitted force through the

individual mounts that are supporting the powertrain. The objective function is described

in Eq. (4.38). The input load vector corresponds to the force is a linear combination of

the shaking force and the road load. The example presented in this section is identical to

the example presented in the previous section. An additional constraint is added to

control the maximum steady state displacement due to the presence of the road load. The

load profile used in this example is shown in Fig. 4.8. The governing equation used for

the V-Twin engine configuration used for the computation of the shaking force and the

road load will be discussed in detail in chapter 5. An elaborate road load model based on
 71

the Pacejka tire model is presented in Appendix B. This model is used to compute the

forces and moments acting on the tire patch. The results of the optimization problem are

presented in Table 4.8 and Table 4.9.

Table 4.8: Optimization Results (12 DOF Model – Combined Loading)

Load Transmitted Mount Stiffness (lb/in)

(lb) x y z

Initial Guess 566.45 475 475 7500

Optimized
122.68 2405 2405 1564.5
Design
1.482 9.087 10.800 11.639 22.275 35.658
Natural
Frequencies 1271.338 (un-
37.437 103.036 103.396 105.657 200.603
damped)
1.41 9.08 10.766 11.621 22.270 35.54
(Hz) 1270.326
37.3068 103.002 103.378 103.628 198.41
(damped)

Damping 0.019 0.021 0.026 0.034 0.040 0.055

Coefficients 0.079 0.081 0.083 0.147 0.195 0.315
 72

Table 4.9: Optimization Results for location and Orientation (12 DOF Model –
Combined Loading)

Mount 1 Mount 2 Mount 3 Mount 4

Orientation (deg)
(0.1, 50, 0) (-0.1, -50, 0) (0.5, 25, 0) (-0.5, -25, 0)
Starting Guess

Results (0.6, 50, 0) (-0.6, -50, 0) (4.9, 0, 0) (-4.9, 0, 0)

Position (in)
(12, -9, 0) (12, -9, 0) (-19, -5, 0) (-19, -5, 0)
Starting Guess

Results (8, -8,-3.4) (8, -8,3.4) (-17, -6.8, -3.2) (-17, -6.8, 3.2)

Road Profile #1

-1
Bump Height (in.)

-2

-3

-4

-5

-6
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Time (sec)

Figure 4.8: Road Profile

 73

1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 1.3719Hz Mode corresponding to 7.7749Hz
1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 20.9711Hz Mode corresponding to 21.793Hz
1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 24.8789Hz Mode corresponding to 26.7216Hz

Figure 4.9: Mode Shapes - 1 to 6 (12 DOF Model – Combined Loading)

1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 62.5727Hz Mode corresponding to 103.126Hz
1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 104.3487Hz Mode corresponding to 201.0303Hz
1 1

0 0

-1 -1
1 2 3 4 5 6 1 2 3 4 5 6
Mode corresponding to 297.2309Hz Mode corresponding to 1271.3498Hz

Figure 4.10: Mode Shapes - 7 to 12 (12 DOF Model – Combined Loading)

 74

4.2.3.3 Discussion of Results

The results for the two optimization problems presented in sections 4.2.3.1 and

4.2.3.2 are shown in Table 4.6, Table 4.7 and Table 4.8, Table 4.9 respectively. Just like

the six DOF model, multiple starting guesses were used due to the nonlinearity of the

mount optimization problem. As can be seen from the tables mentioned above, the out-

of-plane stiffness values are less sensitive to the transmitted loads. The optimum

transmitted load in Tables 4.8 is significantly higher than that shown in Table 4.6. This is

due to the use of a combined loading vector. This loading vector contains the shaking

load and the road load. The in-plane mount stiffness has increased to satisfy the

additional displacement constraint due to the addition of the road load to the input force

vector. The location and orientation vectors show a similar trend to the vectors shown in

the previous section. The optimized mode shapes that correspond to the results shown in

Tables 4.6 and 4.8 are shown in Figs. 4.6, 4.7, 4.9 and 4.10 are normalized. The modes

are numbered from 1 to 12 representing the modes that corresponds to the swing-arm

assembly (modes from 1 to 6), namely (𝑥𝑥𝑠𝑠𝑠𝑠 , 𝑦𝑦𝑠𝑠𝑠𝑠 , 𝑧𝑧𝑠𝑠𝑠𝑠 , 𝛼𝛼𝑠𝑠𝑠𝑠 , 𝛽𝛽𝑠𝑠𝑠𝑠 , 𝛾𝛾𝑠𝑠𝑠𝑠 ), and the modes that

corresponds to the powertrain assembly (modes from 7 to 12), namely

(𝑥𝑥𝑒𝑒 , 𝑦𝑦𝑒𝑒 , 𝑧𝑧𝑒𝑒 , 𝛼𝛼𝑒𝑒 , 𝛽𝛽𝑒𝑒 , 𝛾𝛾𝑒𝑒 ) .

The work done thus far characterizes the engine mount by setting up an objective

function that minimizes the transmitted loads. In certain application, the goal is to design

the mounting system keeping in mind the space limitations where the importance of

decoupling the vibration modes becomes very clear. When the vibration modes are

decoupled the effect of each mode can be examined independently. Although decoupling

vibration modes is not an easy task, a fair bit of work has been done trying to achieve this
 75

goal in the content of vibration isolation. Different approaches have been proposed to

come up with clean decoupled modes for a powertrain mounting system. These methods

include: inclining the isolators and minimizing the off-diagonal terms of the stiffness

matrix.

4.3 Isolator Inclining

The natural modes of vibration can be decoupled through a proper orientation of

the supporting isolators. By doing so, all the modes will exist independently and vibration

of one mode will not excite the other modes. The necessary conditions for decoupling

modes can be stated as follows as mentioned in the vibration and shock handbook

(Harris, 1961). “The resultant of the forces applied to mounted body by the isolators

when the mounted body is displaced in translation must be a force directed through the

center of gravity; or the resultant of the couples applied to the mounted body by the

isolators when the mounted body is displaced in rotation must be a couple about an axis

through the center of gravity”.

Decoupling vibration modes can be achieved by placing the isolators in a plane

that passes through the center of gravity of the powertrain. If this can’t be done,

decoupling can be achieved by inclining the isolators as shown in the below.

 76

𝛽𝛽
z

𝑀𝑀𝑜𝑜 sin 𝜔𝜔𝜔𝜔

𝜖𝜖 φ Elastic Center
x
𝑎𝑎𝑧𝑧
𝑎𝑎𝑧𝑧𝑧𝑧

𝑎𝑎𝑥𝑥 𝑎𝑎𝑥𝑥
𝑘𝑘𝑟𝑟 𝑘𝑘𝑝𝑝

Figure 4.11: Schematic diagram of equipment supported by inclined isolators

If the elastic axis of the mounting system is chosen in a way to pass through the

center of gravity of the powertrain, translational and rotational modes will be decoupled.

Decoupling occurs because the inertia force is being applied through the center of

gravity; as a result the body will not undergo any rotation. To insure complete mode

decoupling, the angle in which the isolators must be inclined must satisfy the following

relation:

1 𝑘𝑘𝑝𝑝
𝑎𝑎𝑧𝑧 2 �1 − �𝑘𝑘𝑟𝑟 �� 𝑠𝑠𝑠𝑠𝑠𝑠2𝜙𝜙
= (4.40)
𝑎𝑎𝑟𝑟 𝑘𝑘𝑞𝑞 𝑘𝑘𝑝𝑝 𝑘𝑘𝑝𝑝
+ + �1 − � �� 𝑠𝑠𝑠𝑠𝑠𝑠2 𝜙𝜙
𝑘𝑘𝑟𝑟 𝑘𝑘𝑟𝑟 𝑘𝑘𝑟𝑟

In Eq. (4.40), k p , k q , k r are the stiffness values along the principal elastic axes of the

isolator and ∅ is the angle between the Z axis and the R axis shown in Fig. 4.11.
 77

4.3.1 Numerical Example

Consider the mounts arranged symmetrically as shown in Fig. 4.12. The mounts

are arranged symmetrically about the z axis. They are attached to one end of the cylinder

at a distance 𝑎𝑎𝑟𝑟 from the z axis and a distance 𝑎𝑎𝑧𝑧 from the x-y plane. The mounts are

inclined so that their principal axes R and P are intersect respectively at two common

points on the z axis. Let the angle between the z axis and the R axis for each mount is ∅.

Let the angle between the z axis and the P axis be 90° − ∅. The Q principal elastic axes

are tangent to the circle of radius 𝑎𝑎𝑟𝑟 which bounds the end face of the cylinder. The mass

m = 200 kg. The inertia values are Ix = Iy = 25 kg.m2 and Iz = 12 kg.m2. The inclination

angle φ = 30o. The stiffness values and the distance are shown in Table 4.10.

x x

Q
P
R
z

𝑎𝑎𝑧𝑧

Figure 4.12: Mount Arrangement

Table 4.10: Stiffness Values and the Distance

kp (N/mm) kq (N/mm) kr (N/mm) az (mm)

50 100 150 123.72

 78

Results for the example presented in this section are shown in Fig. 4.13. The

effect of inclining the mount is clearly seen in Fig. 4:13. This suggests that we need to

find the set of angles that will be used to orient the mount about its axis. This will

decouple the vibration modes. As it’s clearly shown, all of the six modes are completely

decoupled. The natural frequencies of each of the modes are also shown in the Fig. 4.13.

Figure 4.13: Decoupled Modes Along with the Associated Natural Frequency
 79

4.4 Vibration Modes Decoupling

The final approach for mode decoupling considered is formulation of an

optimization problem. The objective function that has been used in this section to achieve

mode decoupling is to minimize the Frobenius norm of the off-diagonal terms of the

overall stiffness matrix. While the value of the objective function is being minimized, the

values of the stiffness, orientation and location of the mounts are being estimated the

SQP optimization technique. The objective function is given by:

𝑛𝑛 𝑛𝑛

𝐹𝐹 = �� � 𝑘𝑘𝑖𝑖𝑖𝑖 � , 𝑖𝑖 ≠ 𝑗𝑗 (4.41)
𝑖𝑖=1 𝑗𝑗 =1
𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓

In Eq. (4.41), 𝑘𝑘𝑖𝑖𝑖𝑖 are the terms of the stiffness matrix. The Frobenius norm is defined as

follows:

𝑇𝑇
𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 = �� 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑�𝐾𝐾𝑜𝑜𝑜𝑜𝑜𝑜 _𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 ∗ 𝐾𝐾𝑜𝑜𝑜𝑜𝑜𝑜 _𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 � (4.42)

To illustrate this procedure a numerical example is presented next where three

different sets of variables are used to achieve mode decoupling.

4.4.1 Numerical Example

A V-6 engine is supported using four mounts. The mass of the engine is 𝑚𝑚 =

276.70 𝑘𝑘𝑘𝑘. The inertia tensor for the engine are shown Table 4.11. The mount

coordinates were measured from the engine center of gravity to each mount attachment

point. The mount’s compression, lateral and force/aft axes define the engine’s x, y and z

coordinate system. Mount orientation is obtained by rotating about the engine x-axis, the

y-axis and the z-axis.

 80

There are three cases through which the approach presented herein is discussed.

The first case corresponds to using the mount stiffness values as the only variables in the

design vector. The second case is where the mount stiffness and orientation form the

design vector. The third and the final case include the mount stiffness values, mount

locations and mount orientation combined together in the design vector. The results for

the three different cases are shown in the Fig. 4.14 through Fig. 4.16. The initial guess

and the final mount parameters along with the matrix showing the minimized off-

diagonal terms of the overall stiffness matrix are shown in Table 4.12 through Table 4.20.

Table 4.111: Inertia Tensor for the Engine

x y z

Ix 15.8 0 0

Iy 0 11.64 0

Iz 0 0 15.69

Table 4.12: Initial Guess of the Mount Locations in (m)

Mount x y z

1 -0.2246 -0.3093 -0.1990

2 0.3614 -0.2823 -0.2510

3 -0.1946 0.1407 -0.2290

4 0.2934 0.1667 -0.2450

 81

Table 4.13: Initial Guess of the Mount Orientations (deg.)

Mount 𝜃𝜃𝑥𝑥 𝜃𝜃𝑦𝑦 𝜃𝜃𝑧𝑧

1 0 -45 0

2 0 -39 180

3 0 -75 0

4 0 -45 180

Table 4.14: Initial Guess of the Mount Stiffness (N/m)

Mount x y z

1 223667 44733 44733

2 170167 126050 48619

3 217167 434334 108583

4 232167 464334 116083

Table 4.15: Optimum Mount Stiffness (Case I)

kx ky kz

1 100 100 36437

2 100 100 26427
3 100 100 77636
4 100 100 46833
 82

Table 4.16: Stiffness Matrix Showing Off-Diagonal Terms (Case I)

4.00E+02 0.00E+00 0.00E+00 0.00E+00 -9.24E+01 2.84E+01

0.00E+00 4.00E+02 0.00E+00 9.24E+01 0.00E+00 2.36E+01

0.00E+00 0.00E+00 1.87E+05 2.92E-04 -9.46E-04 0.00E+00

0.00E+00 9.24E+01 2.92E-04 8.45E+03 2.53E-03 7.33E+00

-9.24E+01 0.00E+00 -9.46E-04 2.53E-03 1.23E+04 -5.93E+00

2.84E+01 2.36E+01 0.00E+00 7.33E+00 -5.93E+00 5.28E+01

Table 4.17: Optimum Mount Stiffness (Case II)

kx ky kz θ1 θ2 θ3

1 253873 36444 10928 -161.81 -57.67 47.57

2 221988 58054 1575 -178.88 55.46 -35.62
3 118594 499001 9737 -106.49 19.36 -78.72
4 135131 410392 12926 108.56 -135.85 169.58

Table 4.18: Stiffness Matrix Showing Off-Diagonal Terms (Case II)

4.73E+05 8.37E-03 -5.32E-03 -1.56E-02 -8.22E-03 2.01E-03

8.37E-03 2.81E+05 1.10E-02 -7.48E-03 5.86E-04 -9.95E-04

-5.32E-03 1.10E-02 1.01E+06 -6.43E-03 1.78E-02 1.50E-02

-1.56E-02 -7.48E-03 -6.43E-03 3.10E+04 2.49E-02 -5.95E-03

-8.22E-03 5.86E-04 1.78E-02 2.49E-02 4.08E+04 3.61E-02

2.01E-03 -9.95E-04 1.50E-02 -5.95E-03 3.61E-02 2.65E+04

 83

Table 4.19: Optimum Mount Stiffness (Case III)

kx ky kz θ1 θ2 θ3 x y z

1 255170 39646 7580 -150.39 47.95 -143.55 -0.32 -0.20 -0.16

2 221410 52800 1005 -173.77 22.53 -43.88 0.46 -0.38 -0.19

3 125570 52542 8132 -153.09 -60.25 -64.77 -0.29 0.24 -0.12

4 133770 408330 12438 104.30 -163.91 173.29 0.19 0.06 -0.24

Table 4.20: Stiffness Matrix Showing Off-Diagonal Terms (Case III)

3.96E+05 -6.74E-02 9.02E-02 1.20E-01 -1.65E-01 -1.09E-01

-6.74E-02 2.78E+05 1.37E-01 1.63E-01 -9.33E-02 -1.22E-01

9.02E-02 1.37E-01 6.45E+05 -8.82E-02 2.16E-02 -2.65E-02

1.20E-01 1.63E-01 -8.82E-02 9.56E+03 -1.40E-01 -4.42E-02

-1.65E-01 -9.33E-02 2.16E-02 -1.40E-01 3.14E+04 9.03E-02

-1.09E-01 -1.22E-01 -2.65E-02 -4.42E-02 9.03E-02 3.28E+04

 84

Figure 4.14: Decoupled Modes and the Corresponding Frequencies (Case I)

Figure 4.15: Decoupled Modes and the Corresponding Frequencies (Case II)
 85

Figure 4.16: Decoupled Modes and the Corresponding Frequencies (Case III)

As can be seen from Fig. 4.14 through Fig. 4.16, the vibration modes are being

decoupled. In the first case were the mount stiffness is the only variable, most of the

modes were decoupled. In the second and third cases, where on top of mount stiffness,

the mount orientations are also considered; all the modes are completely decoupled. This

observation is in line with the previous findings that we have found in the earliear

sections, which emphasize on the importance of mount orientation in achieving

decoupling. These results are seen in Fig. 4.15 and Fig. 4.16 where it can be clearly seen

that all the modes are decoupled nicely.

4.5 Summary

This chapter presents different techniques that are used to characterize the engine

mounting system. All of the techniques presented use the method of Sequential Quadratic

Programming to solve the optimization problem. The SQP method is also introduced in
 86

this chapter as well. The development of two different models of the mounting system is

discussed herein. Both models are presented in detail where the equations of motion are

described in detail. The first model is a simple six DOF powertrain model that is solved

by minimizing the transmitted forces while finding the mount characteristics. The second

model is a more comprehensive model that is used to better understand the vibration

isolation in the motorcycle. The second model is solved in the same way as the first

model. The second model is capable of capturing the effect of the shaking load, the road

load or a combination of both loads. Two more techniques were used to characterize the

mounting system. These techniques are, isolator inclining and vibration modes

decoupling to achieve a complete decoupling of the vibration modes of the system.

 87

5 Chapter 5 – Optimum Design of a Mount System for a V-Twin Engine

This chapter discusses the V-Twin engine configuration that is commonly used in

motorcycle applications. There are two sources of vibration that affect the performance of

a motorcycle engine mount system; the first one is due to the shaking forces which are

generated due to the engine imbalance in the moving parts inside the engine. This force is

transmitted to the frame through the mounting system. The second force is due to the

road loads which are caused by the irregularities in the road profile. These forces are

transmitted to the frame thorough the tire patch. The road load could be periodic or non-

periodic whereas the shaking load is periodic. Numerical examples are presented for

solving the mounting system optimization problem when shaking forces and/or road

loads are present.

5.1 Shaking Loads

This chapter focuses on designing the most suitable mounting system that provides

isolation against forces transmitted from the powertrain to the frame. It is known that

force and motion isolation are the major problems that engineers encounter when

designing an engine mount. Motorcycle engines contain reciprocating parts that produce

shaking forces due to the movement of various parts of the engine. The main objective

herein is to minimize these shaking forces. This objective is achieved by supporting the

powertrain by using a resilient support or an isolator. The largest lumped mass that the

vehicle carries is the powertrain, which is attached to the frame using rubber mounts. The

mounting system that is used in these cases must ensure low vibration transmission

from/into the engine. There are a lot of factors to consider when looking at the source of

vibration, which could be internal or external or both. In this section, attention will
 88

focused on the internal shaking forces which are created due to the engine imbalance

(Paul, 1979). The shaking force is defined as the sum of the inertia and static forces that

are transmitted to the frame through the mounting system (Kaul, 2006). Minimizing the

transmitted loads from the engine to the frame is discussed as well in the literature (Tao,

2000) and (Snyman, 1995) which is considered in detail in Appendix D.

Before developing the expressions for shaking forces in a V-Twin engine, an

analysis will be performed to develop expressions for shaking forces in a single cylinder

engine.

5.1.1 Transmitted Shaking Loads

Figure 5.1 shows a schematic diagram of a single cylinder slider crank

mechanism. The standard slider crank mechanism is the basic building block of virtually

all internal combustion engines. Presented next is the position, velocity, acceleration and

the forces analysis of the slider-crank mechanism. Let the crank radius be r and the

connecting rod length be l. The crank angle is θ and the angle that the connecting rod

makes with the x axis is φ, the crank rotates at a constant speed 𝜔𝜔 then:

𝑞𝑞 = 𝑟𝑟 sin 𝜃𝜃 = 𝑙𝑙 sin 𝜙𝜙 (5.1)

𝜃𝜃 = 𝜔𝜔𝜔𝜔 (5.2)

𝑟𝑟
sin 𝜙𝜙 = sin 𝜔𝜔𝜔𝜔 (5.3)
𝑙𝑙

𝑠𝑠 = 𝑟𝑟 cos 𝜔𝜔𝜔𝜔 and 𝑢𝑢 = 𝑙𝑙 cos 𝜙𝜙 (5.4)

The distance x that is measured from the pivot point O to the slider at point B is

given as follows:

𝑥𝑥 = 𝑠𝑠 + 𝑢𝑢 = 𝑟𝑟 cos 𝜔𝜔𝜔𝜔 + 𝑙𝑙 cos 𝜙𝜙 (5.5)

 89

𝑟𝑟 2
cos 𝜙𝜙 = �1 − 𝑠𝑠𝑠𝑠𝑠𝑠2 𝜙𝜙 = �1 − � sin 𝜔𝜔𝜔𝜔� (5.6)
𝑙𝑙

𝑟𝑟 2
�
𝑥𝑥 = 𝑟𝑟 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 + 𝑙𝑙 1 − � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠� (5.7)
𝑙𝑙

l
r
q

θ φ B x
O

s u

Figure 5.1: Slider Crank Mechanism

The expression given by Eq. (5.7) gives the position of the piston along the x axis

as a function of crank angle θ. If a derivative of Eq. (5.7) is taken once with respect to

time, the velocity of the piston will be determined as shown below:

⎡ ⎤
𝑟𝑟 sin 2𝜔𝜔𝜔𝜔
𝑥𝑥̇ = −𝑟𝑟𝑟𝑟 ⎢sin 𝜔𝜔𝜔𝜔 + ⎥ (5.8)
⎢ 2𝑙𝑙 𝑟𝑟 2⎥
�1 − � sin 𝜔𝜔𝜔𝜔�
⎣ 𝑙𝑙 ⎦
 90

If a derivative of the piston velocity is taken once with respect to time, the piston

acceleration is obtained as shown below:

𝑟𝑟[𝑙𝑙 2 (1 − 2𝑐𝑐𝑐𝑐𝑐𝑐 2 𝜔𝜔𝜔𝜔) − 𝑟𝑟 2 𝑠𝑠𝑠𝑠𝑠𝑠4 𝜔𝜔𝜔𝜔]

𝑥𝑥̈ = −𝑟𝑟𝜔𝜔2 �cos 𝜔𝜔𝜔𝜔 − 3 � (5.9)
[𝑙𝑙 2 − (𝑟𝑟 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠)2 ]2

In the velocity expression shown in Eq. (5.8) and the acceleration expression shown in

Eq. (5.9), a steady state solution is considered where it is assumed that the crank speed ω

is constant.

Using the binomial theorem, an approximate expression for the position, velocity

and acceleration of the piston can be written as follows:

𝑟𝑟 2 𝑟𝑟
𝑥𝑥 ≅ 𝑙𝑙 − + 𝑟𝑟 �cos 𝜔𝜔𝜔𝜔 + cos 2𝜔𝜔𝜔𝜔�
4𝑙𝑙 4𝑙𝑙
𝑟𝑟
𝑥𝑥̇ ≅ −𝑟𝑟𝑟𝑟 �sin 𝜔𝜔𝜔𝜔 + sin 2𝜔𝜔𝜔𝜔�
2𝑙𝑙
𝑟𝑟
𝑥𝑥̈ ≅ −𝑟𝑟𝜔𝜔2 �cos 𝜔𝜔𝜔𝜔 + 𝑙𝑙
cos 2𝜔𝜔𝜔𝜔� (5.10)

The inertia force 𝐹𝐹𝑖𝑖 is the sum of the inertia forces at points A and B on the slider

crank mechanism.

𝐹𝐹𝑖𝑖 = 𝑚𝑚𝐴𝐴 𝑎𝑎𝐴𝐴 + 𝑚𝑚𝐵𝐵 𝑎𝑎𝐵𝐵 (5.11)

In Eq. (5.11), the acceleration term 𝑎𝑎𝐵𝐵 is the acceleration of the piston which is

given in Eq. (5.10). The acceleration term 𝑎𝑎𝐴𝐴 could be found by taking the second

derivative of the position vector at point A with respect to time. The position vector that

describes the location of point A is given as follows:

𝑅𝑅𝐴𝐴 = 𝑟𝑟 cos 𝜔𝜔𝜔𝜔 𝑖𝑖̂ + 𝑟𝑟 sin 𝜔𝜔𝜔𝜔 𝑗𝑗̂ (5.12)

Differentiate the position vector given in Eq. (5.12) twice with respect to time and

an expression for the acceleration at point A is achieved as follows:

 91

𝑎𝑎𝐴𝐴 = [−𝑟𝑟𝑟𝑟 sin 𝜔𝜔𝜔𝜔 − 𝑟𝑟𝜔𝜔2 cos 𝜔𝜔𝜔𝜔] 𝑖𝑖̂ + [𝑟𝑟𝑟𝑟 cos 𝜔𝜔𝜔𝜔 − 𝑟𝑟𝜔𝜔2 sin 𝜔𝜔𝜔𝜔] 𝑗𝑗̂ (5.13)

In Eq. (5.12), 𝑖𝑖̂ and 𝑗𝑗̂ are unit vectors defined along the x and y axis. The inertia force

along the x and y axis are given as follows:

𝑟𝑟 2 𝜔𝜔2
𝐹𝐹𝑖𝑖𝑖𝑖 = −(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 ) 𝑟𝑟𝜔𝜔2 cos 𝜔𝜔𝜔𝜔 − 𝑚𝑚𝐵𝐵 cos 2𝜔𝜔𝜔𝜔
𝑙𝑙

𝑟𝑟 2 𝛼𝛼
− (𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 ) 𝑟𝑟𝑟𝑟 sin 𝜔𝜔𝜔𝜔 − 𝑚𝑚𝐵𝐵 sin 2𝜔𝜔𝜔𝜔
2𝑙𝑙

𝐹𝐹𝑖𝑖𝑖𝑖 = 𝑚𝑚𝐴𝐴 𝑟𝑟𝑟𝑟 cos 𝜔𝜔𝜔𝜔 − 𝑚𝑚𝐴𝐴 𝑟𝑟𝜔𝜔2 sin 𝜔𝜔𝜔𝜔 (5.14)

In Eq. (5.14), mA and mB are the equivalent rotating and reciprocating masses

respectively. The shaking force is 𝐹𝐹𝑠𝑠 = −𝐹𝐹𝑖𝑖 . It is fully described taking into account the

equivalent balancing masses as shown below:

𝑟𝑟 2 𝜔𝜔2
𝐹𝐹𝑠𝑠𝑠𝑠 = (𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 − 𝑚𝑚𝑐𝑐𝑐𝑐 ) 𝑟𝑟𝜔𝜔2 cos 𝜔𝜔𝜔𝜔 + 𝑚𝑚𝐵𝐵 cos 2𝜔𝜔𝜔𝜔
𝑙𝑙

𝑟𝑟 2 𝛼𝛼
+ (𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 − 𝑚𝑚𝑐𝑐𝑐𝑐 ) 𝑟𝑟𝑟𝑟 sin 𝜔𝜔𝜔𝜔 + 𝑚𝑚𝐵𝐵 sin 2𝜔𝜔𝜔𝜔
2𝑙𝑙

𝐹𝐹𝑠𝑠𝑠𝑠 = (𝑚𝑚𝐴𝐴 − 𝑚𝑚𝑐𝑐𝑐𝑐 ) 𝑟𝑟𝜔𝜔2 sin 𝜔𝜔𝜔𝜔 − (𝑚𝑚𝐴𝐴 − 𝑚𝑚𝑐𝑐𝑐𝑐 ) 𝑟𝑟𝑟𝑟 cos 𝜔𝜔𝜔𝜔 (5.15)

In Eq. (5.15), Fsx and Fsy denote the net shaking forces in the x and y directions

respectively and 𝑚𝑚𝑐𝑐𝑐𝑐 is the equivalent mass. These shaking forces result from a single

cylinder.
 92

Presented next is the development of shaking force expressions for a V-twin

engine shown in Fig. 5.2. The shaking force analysis that was done on a single cylinder

engine is generalized to accommodate the V-twin engine and the shaking forces will be

computed along the global X-Y coordinate system.

𝑖𝑖̂

𝑙𝑙̂ 𝑟𝑟̂
𝑛𝑛�

𝑚𝑚
�

l
l

β β

θ
r
𝑗𝑗̂

Figure 5.2: V-twin Engine Configuration.

In order to determine the shaking forces for the V-Twin engine, the shaking force

expression for single cylinder engine given in Eq. (5.15) are used. The forces in each

bank will be computed separately. Then by combining the corresponding terms of the

shaking forces in each bank, the total shaking forces and moments can be computed in

the global X-Y coordinate system for the V-Twin engine.

The shaking force in the left cylinder (bank) (Fs)left is given as follows:
 93

(𝐹𝐹𝑠𝑠 )𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 = �(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 − 𝛽𝛽) − 𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 − 𝛽𝛽)

𝑟𝑟 2 𝜔𝜔2
+ 𝑚𝑚𝐵𝐵 𝑐𝑐𝑐𝑐𝑐𝑐2(𝜃𝜃 − 𝛽𝛽) + (𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝑟𝑟 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽)
𝑙𝑙

𝑟𝑟 2 𝛼𝛼
− 𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝛼𝛼 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽) + 𝑚𝑚𝐵𝐵 𝑠𝑠𝑠𝑠𝑠𝑠2(𝜃𝜃 − 𝛽𝛽)� 𝑙𝑙̂
2𝑙𝑙

+ {𝑚𝑚𝐴𝐴 𝑟𝑟𝜔𝜔2 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽) − 𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝜔𝜔2 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽) − 𝑚𝑚𝐴𝐴 𝑟𝑟𝑟𝑟 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 − 𝛽𝛽)

+ 𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝛼𝛼 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 − 𝛽𝛽)} 𝑚𝑚

� (5.16)

The shaking force in the right cylinder (bank) (Fs)right is given as follows:

(𝐹𝐹𝑠𝑠 )𝑟𝑟𝑟𝑟𝑟𝑟 ℎ𝑡𝑡 = �(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 + 𝛽𝛽) − 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 + 𝛽𝛽)

𝑟𝑟 2 𝜔𝜔2
+ 𝑚𝑚𝐵𝐵 𝑐𝑐𝑐𝑐𝑐𝑐2(𝜃𝜃 + 𝛽𝛽) + (𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝑟𝑟 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽)
𝑙𝑙

𝑟𝑟 2 𝛼𝛼
− 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝛼𝛼 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽) + 𝑚𝑚𝐵𝐵 𝑠𝑠𝑠𝑠𝑠𝑠2(𝜃𝜃 + 𝛽𝛽)� 𝑟𝑟̂
2𝑙𝑙

+ {𝑚𝑚𝐴𝐴 𝑟𝑟𝜔𝜔2 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽) − 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝜔𝜔2 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽) − 𝑚𝑚𝐴𝐴 𝑟𝑟𝑟𝑟 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 + 𝛽𝛽)

+ 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝛼𝛼 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 + 𝛽𝛽)} 𝑛𝑛� (5.17)

In Eq. (5.16) and Eq. (5.17), 𝑟𝑟̂ 𝑎𝑎𝑎𝑎𝑎𝑎 𝑛𝑛� are the unit vectors along the x and y axis of the

local coordinate system for the right cylinder. 𝑙𝑙̂ 𝑎𝑎𝑎𝑎𝑎𝑎 𝑚𝑚

� are the unit vectors along the x

and y local coordinate system for the left cylinder. mcb1 and mcb2 are the equivalent

masses at distances r1 and r2 for the left and right banks respectively. Combining the

shaking forces for the right and left cylinders in their corresponding local coordinate

system and transferring them into the global coordinate system X-Y to come up with the

overall shaking forces of the V-twin engine yields:

 94

𝐹𝐹𝑠𝑠𝑠𝑠 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 �2(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝜔𝜔2 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 − 2(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝑟𝑟 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐

𝑟𝑟 2 𝜔𝜔2 𝑟𝑟 2 𝛼𝛼
+ 2𝑚𝑚𝐵𝐵 𝑠𝑠𝑠𝑠𝑠𝑠2𝜃𝜃𝜃𝜃𝜃𝜃𝜃𝜃2𝛽𝛽 − 𝑚𝑚𝐵𝐵 𝑐𝑐𝑐𝑐𝑐𝑐2𝜃𝜃𝜃𝜃𝜃𝜃𝜃𝜃2𝛽𝛽 �
𝑙𝑙 𝑙𝑙

+ 𝜔𝜔2 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 {𝑚𝑚𝑐𝑐𝑐𝑐2 𝑟𝑟2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 + 𝛽𝛽) − 𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 − 𝛽𝛽)}

+ 𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼 {𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽) − 𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽)}

+ 𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽 {2𝑚𝑚𝐴𝐴 𝑟𝑟𝑟𝑟2 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 − 2𝑚𝑚𝐴𝐴 𝑟𝑟𝑟𝑟 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐}

+ 𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼 {𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 − 𝛽𝛽) + 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 + 𝛽𝛽)}

− 𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 {𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽) + 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽) } (5.18)

𝐹𝐹𝑠𝑠𝑠𝑠 = 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 �2(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 + 2(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝑟𝑟 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠

𝑟𝑟 2 𝜔𝜔2 𝑟𝑟 2 𝛼𝛼
+ 2𝑚𝑚𝐵𝐵 𝑐𝑐𝑐𝑐𝑐𝑐2𝜃𝜃𝜃𝜃𝜃𝜃𝜃𝜃2𝛽𝛽 + 𝑚𝑚𝐵𝐵 𝑠𝑠𝑠𝑠𝑠𝑠2𝜃𝜃𝜃𝜃𝜃𝜃𝜃𝜃2𝛽𝛽 �
𝑙𝑙 𝑙𝑙

− 𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 {𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 − 𝛽𝛽) + 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 + 𝛽𝛽)}

− 𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼 {𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽) + 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽)}

+ 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 {2𝑚𝑚𝐴𝐴 𝑟𝑟𝑟𝑟2 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 + 2𝑚𝑚𝐴𝐴 𝑟𝑟𝑟𝑟 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠}

+ 𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼 {𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 + 𝛽𝛽) − 𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 − 𝛽𝛽)}

+ 𝜔𝜔2 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 {𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽) − 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽) } (5.19)

The shaking forces shown in Eq. (5.18) and Eq. (5.19) can be employed to find

the shaking moments by multiplying each term by the moment arm. The moments exist

within each bank and their vectors will be orthogonal to the cylinder planes. For the right

bank, a moment unit vector 𝑛𝑛� is defined which is perpendicular to the unit vector 𝑟𝑟̂ .
 95

� is defined which is perpendicular to the unit vector 𝑙𝑙̂

Similarly, a moment unit vector 𝑚𝑚

for the left bank as shown in Fig. 5.2.

The shaking moment in the left cylinder (bank) (Ms)left is given as follows:

(𝑀𝑀𝑠𝑠 )𝑙𝑙𝑙𝑙𝑓𝑓𝑓𝑓 = �(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 − 𝛽𝛽) − 𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 − 𝛽𝛽)

𝑟𝑟 2 𝜔𝜔2
+ 𝑚𝑚𝐵𝐵 𝑐𝑐𝑐𝑐𝑐𝑐2(𝜃𝜃 − 𝛽𝛽) + (𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝑟𝑟 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽)
𝑙𝑙

𝑟𝑟 2 𝛼𝛼
− 𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝛼𝛼 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽) + 𝑚𝑚𝐵𝐵 𝑠𝑠𝑠𝑠𝑠𝑠2(𝜃𝜃 − 𝛽𝛽)� 𝑧𝑧 𝑚𝑚
� (5.20)
2𝑙𝑙

The shaking moment in the right cylinder (bank) (Ms)right is given as follows:

(𝑀𝑀𝑠𝑠 )𝑟𝑟𝑟𝑟𝑟𝑟 ℎ𝑡𝑡 = �(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 + 𝛽𝛽) – 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 + 𝛽𝛽)

𝑟𝑟 2 𝜔𝜔2
+ 𝑚𝑚𝐵𝐵 𝑐𝑐𝑐𝑐𝑐𝑐2(𝜃𝜃 + 𝛽𝛽) + (𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝑟𝑟 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽)
𝑙𝑙

𝑟𝑟 2 𝛼𝛼
− 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝛼𝛼 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽) + 𝑚𝑚𝐵𝐵 𝑠𝑠𝑠𝑠𝑠𝑠2(𝜃𝜃 + 𝛽𝛽)� 𝑧𝑧 𝑛𝑛� (5.21)
2𝑙𝑙

In Eq. (5.20) and Eq. (5.21), z is the moment arm. Combining the shaking moments for

the right and left cylinders that have been shown Eq. (5.20) and Eq. (5.21) in their

corresponding local coordinate system and transferring them into the global coordinate

system X-Y to come up with the overall shaking moments for the V-twin engine yields:

𝑟𝑟 2 𝜔𝜔2
𝑀𝑀𝑠𝑠𝑠𝑠 = 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 �2(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 + 2𝑚𝑚𝐵𝐵 𝑐𝑐𝑐𝑐𝑐𝑐2𝜃𝜃𝜃𝜃𝜃𝜃𝜃𝜃2𝛽𝛽
𝑙𝑙

𝑟𝑟 2 𝛼𝛼
+ 2(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝑟𝑟 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 + 𝑚𝑚𝐵𝐵 𝑠𝑠𝑠𝑠𝑠𝑠2𝜃𝜃𝜃𝜃𝜃𝜃𝜃𝜃2𝛽𝛽 � . 𝑧𝑧
𝑙𝑙

− 𝜔𝜔2 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 {𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 − 𝛽𝛽) + 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 + 𝛽𝛽)} . 𝑧𝑧

− 𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼 {𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽) + 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽)} . 𝑧𝑧 (5.22)
 96

𝑟𝑟 2 𝜔𝜔2
𝑀𝑀𝑠𝑠𝑠𝑠 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 �−2(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝜔𝜔2 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 − 2𝑚𝑚𝐵𝐵 𝑠𝑠𝑠𝑠𝑠𝑠2𝜃𝜃𝜃𝜃𝜃𝜃𝜃𝜃2𝛽𝛽
𝑙𝑙

𝑟𝑟 2 𝛼𝛼
+ 2(𝑚𝑚𝐴𝐴 + 𝑚𝑚𝐵𝐵 )𝑟𝑟𝑟𝑟 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 + 𝑚𝑚𝐵𝐵 𝑐𝑐𝑐𝑐𝑐𝑐2𝜃𝜃𝜃𝜃𝜃𝜃𝜃𝜃2𝛽𝛽 � . 𝑧𝑧
𝑙𝑙

+ 𝜔𝜔2 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 {𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 − 𝛽𝛽) − 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃 + 𝛽𝛽)} . 𝑧𝑧

+ 𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼𝛼 {𝑚𝑚𝑐𝑐𝑐𝑐 1 𝑟𝑟1 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽) − 𝑚𝑚𝑐𝑐𝑐𝑐 2 𝑟𝑟2 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽)} . 𝑧𝑧 (5.23)

The shaking torque or one cylinder is calculated using the inertia force acting on

the piston Fi14 multiplied by the distance x from the piston at point B to the origin of the

coordinate system at point O as shown in Fig. 5.1. The free body diagram of the piston

showing all the acting forces are shown below in Fig. 5.3.

𝐹𝐹𝑖𝑖34
𝐹𝐹𝑖𝑖34
B 𝐹𝐹𝑖𝑖14
𝑚𝑚𝐵𝐵 𝑥𝑥̈
φ

𝑚𝑚𝐵𝐵 𝑥𝑥̈

𝐹𝐹𝑖𝑖14

Figure 5.3: Free Body Diagram of the Piston

𝑇𝑇𝑠𝑠 = (𝐹𝐹𝑖𝑖14 ∗ 𝑥𝑥)𝑘𝑘�

= 𝑚𝑚𝐵𝐵 𝑥𝑥̈ 𝑡𝑡𝑡𝑡𝑡𝑡∅ ∗ 𝑥𝑥 (5.24)

In Eq. (5.24), 𝑥𝑥̈ is the piston acceleration represented in Eq. (5.10). Substitute the piston

acceleration expressed in Eq. (5.10) into Eq. (5.24), the shaking torque will be expressed

as follows:
 97

𝑟𝑟 𝑟𝑟
𝑇𝑇𝑠𝑠 = 𝑚𝑚𝐵𝐵 �−𝑟𝑟𝜔𝜔2 �𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 + 𝑐𝑐𝑐𝑐𝑐𝑐2𝜔𝜔𝜔𝜔� − 𝑟𝑟𝑟𝑟 �𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 + 𝑠𝑠𝑠𝑠𝑠𝑠2𝜔𝜔𝜔𝜔�� 𝑡𝑡𝑡𝑡𝑡𝑡∅ ∗ 𝑙𝑙
𝑙𝑙 2𝑙𝑙

𝑟𝑟 2 𝑟𝑟
− + 𝑟𝑟 �𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 + 𝑐𝑐𝑐𝑐𝑐𝑐2𝜔𝜔𝜔𝜔� (5.25)
4𝑙𝑙 4𝑙𝑙

𝑟𝑟 𝑟𝑟 2
𝑡𝑡𝑡𝑡𝑡𝑡∅ ≈ 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 �1 + 2 𝑠𝑠𝑠𝑠𝑠𝑠2 𝜔𝜔𝜔𝜔� (5.26)
𝑙𝑙 2𝑙𝑙

In Eq. (5.25), x is described in Eq. (5.10) and 𝑘𝑘� is unit vector acting along the z axis

which is perpendicular to the plane of the slider crank mechanism shown in Fig. 5.1.

Assume that the angular acceleration α is zero and approximating tan φ as shown in Eq.

(5.26) we get an expression for the shaking torque for both the left and the right banks.

The shaking torque in the left bank (Ts)left is as follows:

1 𝑟𝑟
(𝑇𝑇𝑠𝑠 )𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 = 𝑚𝑚𝐵𝐵 𝑟𝑟 2 𝜔𝜔2 � 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 − 𝛽𝛽) − 𝑠𝑠𝑠𝑠𝑠𝑠2(𝜃𝜃 − 𝛽𝛽)
2 2𝑙𝑙

3𝑟𝑟
− 𝑠𝑠𝑠𝑠𝑠𝑠3(𝜃𝜃 − 𝛽𝛽)� 𝑘𝑘� (5.27)
2𝑙𝑙

The shaking torque in the right bank (Ts)right is as follows:

1 𝑟𝑟
(𝑇𝑇𝑠𝑠 )𝑟𝑟𝑟𝑟𝑟𝑟 ℎ𝑡𝑡 = 𝑚𝑚𝐵𝐵 𝑟𝑟 2 𝜔𝜔2 � 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃 + 𝛽𝛽) − 𝑠𝑠𝑠𝑠𝑠𝑠2(𝜃𝜃 + 𝛽𝛽)
2 2𝑙𝑙

3𝑟𝑟
− 𝑠𝑠𝑠𝑠𝑠𝑠3(𝜃𝜃 + 𝛽𝛽)� 𝑘𝑘� (5.28)
2𝑙𝑙

The combined shaking torque Ts due to shaking torque from both left and right banks is

the algebraic sum of both (Ts)left and (Ts)right shown below:

1 𝑟𝑟 3𝑟𝑟
𝑇𝑇𝑠𝑠 = 𝑚𝑚𝐵𝐵 𝑟𝑟 2 𝜔𝜔2 � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 − 2𝑠𝑠𝑠𝑠𝑠𝑠2𝜃𝜃𝜃𝜃𝜃𝜃𝜃𝜃2𝛽𝛽 − 𝑠𝑠𝑠𝑠𝑠𝑠3𝜃𝜃𝜃𝜃𝜃𝜃𝜃𝜃3𝛽𝛽� 𝑘𝑘� (5.29)
2 𝑙𝑙 𝑙𝑙

The final shaking force vector is a 6x1 vector shown below:

 98

𝑇𝑇
𝐹𝐹𝑠𝑠 = �𝐹𝐹𝑠𝑠𝑠𝑠 𝐹𝐹𝑠𝑠𝑠𝑠 𝐹𝐹𝑠𝑠𝑠𝑠 𝑀𝑀𝑠𝑠𝑠𝑠 𝑀𝑀𝑠𝑠𝑠𝑠 𝑇𝑇𝑠𝑠 � (5.30)

5.1.2 Numerical Example

The example discussed herein is based on the six DOF model presented in section

3.4.1 and the optimization problem of minimizing the transmitted load was formulated in

section 4.2. This example is a continuation of the example presented in section 4.2.2.1

where the shaking force is the only input load. The input load is calculated at different

steady speeds of 800 rpm, 3000 rpm and 5000 rpm. The powertrain is supported by four

identical circular cross section elastomeric mounts. The layout of this system is shown in

Fig. 4.4, and the objective function is presented in Eq. (4.38). Mount parameters, which

consists of mount stiffness, mount locations and mount orientations are compiled to form

the design vector. The powertrain mass and inertia tensor and the lower and upper bounds

used for the design variables and the limit that is imposed on the design variables by

constraining the deflection of the powertrain are the same as those imposed in the

example presented in section 4.2.2.1.

The optimization problem is solved using the SQP technique that employs a

function (fmincon) to minimize the value of the objective function. The force vector

corresponding to different engine steady speeds is shown in Table 5.1. The design

variables resulting from the optimization process are shown in Table 5.2 to Table 5.4.

The resulting force plots in the x and y directions for different engine speeds are shown in

Fig. 5.4 to Fig. 5.6 and the resulting torque plots for different engine speeds are shown in

Fig. 5.7 to Fig. 5.9.

 99

Shaking Force corresponding to 800 rpm

400
Fore-aft (x)
Vertical (y)
300

Shaking force (lb) 200

100

-100

-200

-300
0 100 200 300 400 500 600 700 800
Crank angle (deg)

Figure 5.4: Shaking Force in the x and y Directions (800 rpm)

Shaking Force corresponding to 3000 rpm

5000
Fore-aft (x)
4000 Vertical (y)

3000

2000
Shaking force (lb)

1000

-1000

-2000

-3000

-4000
0 100 200 300 400 500 600 700 800
Crank angle (deg)

Figure 5.5: Shaking Force in the x and y Directions (3000 rpm)

 100

4
x 10 Shaking Force corresponding to 5000 rpm
1.5
Fore-aft (x)
Vertical (y)

1
Shaking force (lb)

0.5

-0.5

-1
0 100 200 300 400 500 600 700 800
Crank angle (deg)

Figure 5.6: Shaking Force in the x and y Directions (5000 rpm)

Shaking torque corresponding to 800 rpm

15

10

5
Torque (lb-ft)

-5

-10

-15
0 100 200 300 400 500 600 700 800
Crank angle (deg)

Figure 5.7: Shaking Torque (800 rpm)

 101

Shaking torque corresponding to 3000 rpm

200

150

100

50
Torque (lb-ft)

-50

-100

-150

-200
0 100 200 300 400 500 600 700 800
Crank angle (deg)

Figure 5.8: Shaking Torque (3000 rpm)

Shaking torque corresponding to 5000 rpm

500

400

300

200

100
Torque (lb-ft)

-100

-200

-300

-400

-500
0 100 200 300 400 500 600 700 800
Crank angle (deg)

Figure 5.9: Shaking Torque (5000 rpm)

 102

Table 5.1: Shaking Force Vector at Different Speeds

ω Fx Fy Fz Mx My T

(rpm) (lb) (lb.ft)

800 131.12 308.82 0 0 0 12.15

3000 1843.87 4342.79 0 0 0 170.89

5000 5121.87 12063.3 0 0 0 474.70

Table 5.2: Optimization Results for the Shaking Force Vector Corresponding to 800 rpm

Load
Mount Stiffness (lb/in)
Transmitted
(lb) x y z

Initial Guess 412.48 475 475 7500

Optimized Design 188.81 5000 5000 7585.3

Table 5.3: Optimization Results for the Shaking Force Vector Corresponding to 3000
rpm

Load
Mount Stiffness (lb/in)
Transmitted
(lb) x y z

Initial Guess 815.49 475 475 7500

Optimized Design 247.18 1728 1728 7993.6

 103

Table 5.4: Optimization Results for the Shaking Force Vector Corresponding to 5000
rpm

Load
Mount Stiffness (lb/in)
Transmitted
(lb) x y z

Initial Guess 448.19 475 475 7500

Optimized Design 259.83 1278.2 1278.2 7515.7

As can be seen from the plots shown above, the forces in the x and y directions as

well as the shaking torque exhibit a periodic behavior. It is clear from Table 5.1 that the

value of the forces and moments increase with the increase of the engine operating speed.

Table 5.2 through Table 5.4 shows the optimum stiffness values for the mounting system.

The results are consistent with the results found in section 4.2.2.1 where the stiffness

along the z-axis was seen to have no influence on the objective function value and thus

did not deviate much from the starting guess value whereas the stiffness values in the x

and y direction do change significantly. This is due to the fact the force transmitted to the

frame is less sensitive to the stiffness in the z-direction.

5.2 Road Loads

In this section, the effect of external loads on the mounting system is investigated.

One of the main problems that engineers encounter in vibration isolation is the problem

of motion isolation. This problem is seen in the case of external loads that are transmitted

to the engine. These loads which are due to the irregularities of the road profile are

transmitted to the frame through the tire patch. Two different road profiles are

investigated in this work with one road profile which is periodic while the other road
 104

profile is non-periodic. The main goal is to come up with an appropriate mounting system

that minimizes the transmission of these external loads.

5.2.1 Transmitted Road Loads

The road loads are due to irregularities in the road profile which could be periodic

or non-periodic. These road profiles are analyzed for a specific displacement functions in

which the frequency content is analyzed. For the periodic profiles, the frequency content

is obtained by using the Fourier series expansion of the displacement function. For the

non-periodic profiles, it’s obtained using the Fourier transform. Herein, the Fourier series

coefficient and the frequency content are obtained using the Fast Fourier Transforms

(FFT) (Chen, 2001). The input force resulting from a certain road profile is determined as

follows:

𝐹𝐹𝑦𝑦 = 𝑘𝑘𝑘𝑘 + 𝑐𝑐𝑥𝑥̇ (5.31)

In Eq. (5.31), Fy is the vertical component of the force that is transmitted through the tire

patch due to the displacement x and the velocity 𝑥𝑥̇ as a result of the road profile change. k

and c are the stiffness and damping of the rear wheel in the y-direction.

The continuous time Fourier series (CTFS) for a periodic road profile is

represented as follows:
∞
2𝜋𝜋
𝑥𝑥(𝑡𝑡) = � 𝑐𝑐𝑚𝑚 𝑒𝑒 𝑗𝑗𝑗𝑗 𝜔𝜔 𝑜𝑜 𝑡𝑡 ; where 𝜔𝜔𝑜𝑜 = (5.32)
𝑃𝑃
𝑚𝑚 = −∞

In Eq. (5.32), cm represents the Fourier series coefficients and are determined as follows:
𝑃𝑃�
2
1
𝑐𝑐𝑚𝑚 = � 𝑥𝑥(𝑡𝑡) 𝑒𝑒 −𝑗𝑗𝑗𝑗 𝜔𝜔 𝑜𝑜 𝑡𝑡 𝑑𝑑𝑑𝑑 (5.33)
𝑃𝑃
−𝑃𝑃�
2
 105

In Eq. (5.33), P is the fundamental period of the displacement function x(t) that

corresponds to the fundamental frequency ωo. On the other hand, the discrete time

Fourier series (DTFS) is represented for the discrete displacement function as follows:

𝑥𝑥[𝑛𝑛] = 𝑥𝑥(𝑛𝑛𝑛𝑛) = � 𝑐𝑐𝑚𝑚𝑚𝑚 𝑒𝑒 𝑗𝑗𝑗𝑗 𝜔𝜔 𝑜𝑜 𝑛𝑛𝑛𝑛 (5.34)

𝑚𝑚 = <𝑁𝑁>

where 𝜔𝜔𝑜𝑜 = 2𝜋𝜋/𝑁𝑁𝑁𝑁 and cmd are the Fourier series coefficients which are determined as

follows:
𝑁𝑁−1
1
𝑐𝑐𝑚𝑚𝑚𝑚 = � 𝑥𝑥[𝑛𝑛]𝑒𝑒 −𝑗𝑗𝑗𝑗 𝜔𝜔 𝑜𝑜 𝑛𝑛𝑛𝑛 (5.35)
𝑁𝑁
𝑛𝑛 =0

In Eq. (5.34) and Eq. (5.35), ωo is the fundamental frequency and T is the sampling

period. The DTFS coefficients can be determined using Eq. (5.36), if the band limited

displacement function x(t) and an appropriate sampling period T is chosen using FFT.

𝑋𝑋[𝑚𝑚]
𝑐𝑐𝑚𝑚𝑚𝑚 = (5.36)
𝑁𝑁

In Eq. (5.36), X[m] is the FFT of x[n] and N is the number of terms of x[n] used to

compute the FFT.

The continuous time Fourier transform (CTFT) of the displacement function is

given in Eq. (5.37) and the Discrete time Fourier transform (DTFT) is given in Eq. (5.38)

below.
∞

𝑋𝑋(𝜔𝜔) = � 𝑥𝑥(𝑡𝑡)𝑒𝑒 −𝑗𝑗𝑗𝑗𝑗𝑗 𝑑𝑑𝑑𝑑 (5.37)

−∞

𝑋𝑋𝑑𝑑 (𝜔𝜔) = � 𝑥𝑥(𝑛𝑛𝑛𝑛)𝑒𝑒 −𝑗𝑗𝑗𝑗𝑗𝑗𝑗𝑗 (5.38)

𝑛𝑛=−∞
 106

In Eq. (5.37) and Eq. (5.38), X(ω) is the spectrum of x(t) which can used for periodic and

non-periodic displacement functions.

5.2.2 Numerical Example

The example presented herein discusses the presence of both the shaking loads

and the road loads. The input load vector is a linear combination of the shaking loads and

the road loads in which the shaking loads are evaluated at a steady engine speed of 4000

rpm. The example is a continuation for the example presented in section 4.2.3.2. The

optimization problem is formulated such that the mount parameters which consist of

mount stiffness, location and orientation are compiled to form the vector of design

variables. The layout of the problem considered in this example is shown in Fig. 4.4 and

all the data are presented in section 4.2.3.2.

The optimization problem is solved using the SQP technique that employs a

function (fmincon) to minimize the value of the objective function. Two different road

profiles examined in this example separately in which one is periodic (Profile #1) and the

other one is non-periodic (Profile #2). Both of the road profiles along with their

displacement functions and their corresponding magnitude spectrum and their

reconstructed plots are presented in Fig. 5.10 to Fig. 5.15.

 107

Road Profile #1

-1

Bump Height (in.)

-2

-3

-4

-5

-6
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Time (sec)

Figure 5.10: Road Profile #1

Magnitude plot of the Spectrum for Road Profile #1

0.12

0.1

0.08

0.06

0.04

0.02

0
-6 -4 -2 0 2 4 6
Frequency (rad/sec) 4
x 10
Figure 5.11: Magnitude Plot of the Spectrum for Road Profile #1
 108

Reconstructed Time Plot from Fourier Series Coefficients for Road Profile #1
1

-1

-2

-3

-4

-5

-6
0 0.01 0.02 0.03 0.04 0.05 0.06
Time

Figure 5.12: Reconstructed Time Plot for Road Profile #1

Figure 5.13 shows road profile #2. This road profile is non-periodic with bump

height of 3.5 in. The time displacement plot is shown in Fig. 5.13. The magnitude plot of

the spectrum and the reconstructed plot are shown in Fig. 5.14 and Fig. 5.15 respectively.
 109

Road Profile #2
4

3.5

2.5
Bump Height (in.)

1.5

0.5

-0.5
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (sec)

Figure 5.13: Road Profile #2

Magnitude Plot of the Spectrum for Road Profile #2

4.5

3.5

2.5

1.5

0.5

0
-800 -600 -400 -200 0 200 400 600 800
Frequency (rad/sec)

Figure 5.14: Magnitude Plot of the Spectrum of Road Profile #2

 110

Reconstructed Time Plot from Fourier Series Coefficients for Road Profile #2
4

3.5

2.5

1.5

0.5

-0.5
0 0.5 1 1.5 2 2.5

Figure 5.15: Reconstructed Time Plot for Road Profile #2

The design variables resulting from the optimization process are shown in Table

5.5 and Table 5.7, meanwhile the optimum mount location and orientation for both load

profiles are shown in Table 5.6 and Table 5.8.

Table 5.5: Optimization Results for Road Profile #1

Load
Mount Stiffness (lb/in)
Transmitted
(lb) x y z

Initial Guess 110.40 475 475 2400

Optimized Design 60.01 2336.5 2336.5 2344.8

 111

Table 5.6: Optimization Results for location and Orientation (Profile #1)

Mount 1 Mount 2 Mount 3 Mount 4

Orientation (deg)
(0.1, 50, 0) (-0.1, -50, 0) (0.5, 25, 0) (-0.5, -25, 0)
Starting Guess

Results (0, 50, 0) (0, -50, 0) (50,50, 0) (-50,-50, 0)

Position (in)
(12, -9, 0) (12, -9, 0) (-19, -5, 0) (-19, -5, 0)
Starting Guess

Results (12, -8.9, -3) (12, -8.9, 3) (-17, -10, -7) (-17, -10, 7)

Table 5.7: : Optimization Results for Road Profile #2

Load
Mount Stiffness (lb/in)
Transmitted
(lb) x Y z

Initial Guess 599.63 475 475 7500

Optimized Design 113.47 2116.2 2116.2 5884.5

 112

Table 5.8: Optimization Results for location and Orientation (Profile #2)

Mount 1 Mount 2 Mount 3 Mount 4

Orientation (deg)
(0.1, 50, 0) (-0.1, -50, 0) (0.5, 25, 0) (-0.5, -25, 0)
Starting Guess

Results (50, 50, 0) (-50, -50, 0) (50, 50, 0) (-50, -50, 0)

Position (in)
(12, -9, 0) (12, -9, 0) (-19, -5, 0) (-19, -5, 0)
Starting Guess

Results (12, -5.5,-7) (12, -5.5,7) (-11, -10, -3) (-11, -10, 3)

As can be seen from the tables above, the value of the load transmitted for both

cases is larger than the transmitted loads in the corresponding example discussed in

section 4.3.2.1 where only the shaking force is considered. This due to the fact that the

input force vector contains alongside the shacking load at 4000 rpm the road load due to

different road profiles. It is worth mentioning that different starting guess vector is used

to insure that the final solution is not at local minima. Again, the influence of mount

orientation is seen to be more significant than mount location. The results are consistent

with the results found in section 4.2.2.1 where the stiffness along the z axis does not

affect objective function values and therefore does deviate much from the starting guess

values whereas the stiffness values in the x and y direction do change significantly. An

alternate model is used to estimate the forces and moments transmitted through the tire

patch using the Pacejka model discussed in Appendix B. This tire model is not used in

this dissertation.
 113

5.3 Summary

In this chapter, the input load vector acting on the mounting system is discussed.

The load vector is a linear combination of the shaking force and the road loads. The

shaking force vector is developed first for a single cylinder engine and then generalized

to accommodate a V-Twin engine configuration. The road loads which are due to

irregularities in the road profile are also discussed. Two different road profiles are

considered. Road profile #1, which is periodic and road profile #2, which is non-periodic

are employed in this work. The frequency content of the road profiles is obtained using

the FFT technique. The technique used herein could be used for any other load profile. It

is seen that for both examples considered herein significant reduction in the loads

transmitted through the mount are obtained by varying the mount stiffness values,

location and orientation.

 114

6 Chapter 6 – Shape Optimization of Engine Mounts

In this chapter, a parametric approach is presented to determine the optimum

geometric shape of an engine mount in order to minimize the vibrations transmitted to

and from the engine. The engine mount used in this chapter is an elastomeric mount

which is made of rubber. For proper vibration isolation, elastomeric mounts are designed

such that they have the necessary stiffness rate in all directions. As shown in chapters 4

and 5, an optimization problem is solved first to determine the optimum values of

stiffness, orientation and location of the mount system such that vibrations transmitted

are minimized. Besides knowing the mount stiffness values, a determination of the

optimum shape of the mount is also vital. This chapter addresses determining the shape

of the mount such that it meets the required stiffness of the mounting system obtained

from dynamic analysis. A nonlinear finite element analysis is used to determine the final

optimum shape corresponding to the desired mount stiffness values.

6.1 Finite Element Modeling

The general shape of the shear rubber engine mount, also known as bush engine

mount, is shown in Fig. 6.1. This type of engine mount is common in motorcycle

applications. The finite element model of the mount used herein is created using the finite

element software ANSYS® and is built to fully describe the geometry of the bush type

engine mount. The finite element model is built using solid 186 elements shown in Fig.

6.2. This element is a three dimensional, 20 node structural solid element. The solid

element exhibits quadratic displacement behavior. Each node has a 3 DOF, namely

translation in the x, y and z directions. The element supports plasticity, hyperelasticity,

stress stiffening and large deformations.

 115

The mount is connected to the frame via metal steel plates on both sides. These

plates are bonded to the mount and the connection is at the mount attachment holes. Since

the stiffness of the steel plates is higher than the mount stiffness, the constraints are

moved from the plate holes directly into the mount surface as shown in Figs. 6.3 and 6.4.

The boundary conditions are applied by constraining the displacement of the surface of

the mount in all directions. A shaft which is connected to the source of vibration runs

through the mount which transmits the loads from the powertrain to the frame. Since load

transmitting shaft runs through the center hole of the mount, the loading can be modeled

by using rigid bodies that runs from the center node into the inner mount surface or by

defining a constraint equation as shown in Figs. 6.5 and 6.6. The input load, which is

determined from the dynamic analysis performed in chapters 4 and 5, can be defined by

applying a force at the center node in the x, y and z directions.

Figure 6.1: Mount Geometry

 116

Figure 6.2: Solid 186 Element (ANSYS, 2009)

Figure 6.3: Isometric View Showing the Boundary Conditions

 117

Figure 6.4: Front View Showing the Boundary Conditions

Figure 6.5: Isometric View Showing the Constraint Equation

 118

Figure 6.6: Front View Showing the Constraint Equation

6.2 Parameter Optimization

A parametric study is employed herein to determine the optimum geometric

dimensions of the isomeric mount. These dimensions could be selected to describe the

entire mount as shown in Fig. 6.1. The shape of the mount can be determined by

matching the stiffness that is determined from the dynamic analysis which is performed

in chapter 4 and 5 and the stiffness obtained from the geometry of the mount. Once the

design parameters are chosen which are 𝑡𝑡𝑟𝑟 , 𝑡𝑡𝑠𝑠 , 𝑡𝑡𝑧𝑧 , and 𝜃𝜃, the objective function is set up as

follows:
2
𝜓𝜓 = 𝑤𝑤𝑤𝑤(1)(𝑘𝑘𝑥𝑥 − 𝑘𝑘𝑥𝑥𝑑𝑑𝑑𝑑𝑑𝑑 )2 + 𝑤𝑤𝑤𝑤(2)�𝑘𝑘𝑦𝑦 − 𝑘𝑘𝑦𝑦𝑑𝑑𝑑𝑑𝑑𝑑 � + 𝑤𝑤𝑤𝑤(3)(𝑘𝑘𝑧𝑧 − 𝑘𝑘𝑧𝑧𝑑𝑑𝑑𝑑𝑑𝑑 )2 (6.1)

The bounds of the design variables must satisfy the condition described in Eq.

(6.2), where xi is the ith design variable and n is the number of design variables.

𝑥𝑥𝑖𝑖𝑚𝑚𝑚𝑚𝑚𝑚 ≤ 𝑥𝑥𝑖𝑖 ≤ 𝑥𝑥𝑖𝑖𝑚𝑚𝑚𝑚𝑚𝑚 𝑓𝑓𝑓𝑓𝑓𝑓 𝑖𝑖 = 1, … , 𝑛𝑛 (6.2)

 119

In Eq. (6.1), 𝑤𝑤𝑤𝑤(𝑖𝑖) is the weighting function that corresponds to the stiffness in

the ith direction. The superscript ‘des’ indicates the desired stiffness that is obtained from

the dynamic analysis of the mounting system. Meanwhile the design parameters selected

will determine the stiffness values for the geometry that is obtained from the nonlinear

finite element analysis. The process of determining the design variables is expensive and

time consuming, therefore in order to reduce the number of function evaluations, the least

effective stiffness could be dropped from the objective function 𝜓𝜓.

6.3 Design Model and Analysis

The mount that is used herein is a bush type which is typically used in the

automotive industry. The actual geometry of such mount is shown in Fig. 6.1 along with

the parameters that define its shape. There are a total of six parameters that dictates the

shape of the mount of which four are used as the design variables; namely 𝑡𝑡𝑠𝑠 , 𝑡𝑡𝑧𝑧 , 𝑡𝑡𝑟𝑟 and 𝜃𝜃.

The other two parameters (𝑟𝑟𝑖𝑖 and 𝑟𝑟𝑜𝑜 ) are constants. These design variables affect the

mount stiffness directly. The weighting function that is used in the objective function

could be used to take into account the importance of the stiffness in a particular direction.

The dynamic analysis is done for a motorcycle powertrain in which is supported by four

isomeric mounts. The connection between the powertrain and the swing-arm are taken

into consideration generating a twelve DOF system (Kaul, 2006). The exciting force will

be a mix of the shaking force evaluated at different engine steady state speeds and/or the

road loads.

In the work presented in this chapter, the stiffness values corresponding to

specific mount geometry are obtained using a nonlinear finite element analysis. The

geometry shown in Fig. 6.1 is used to generate a mesh for the analysis. The optimization
 120

is carried out using ANSYS®. Solid 186 is the element that has been used for this purpose

is shown in Fig. 6.2. Appropriate boundary conditions have been applied to the model as

discussed in section 6.2. The boundary conditions along with the constraint equation are

shown in Fig. 6.3 to Fig. 6.6. This model is assumed to exhibit small deflections, for this

reason the Mooney Rivlin model is sufficient to describe the fully incompressible

hyperelastic material behavior of rubber (Kim, 1997), (Rivlin, 1992) and (Ali, 2010). The

Mooney Rivlin model of the strain energy is expressed as follows:

𝑈𝑈 = 𝐶𝐶10 (𝐼𝐼1 − 3) + 𝐶𝐶01 (𝐼𝐼2 − 3) (6.3)

In Eq. (6.2), I1 and I2 are the first and second strain invariants. The coefficients C10 and

C01 are determined from the uniaxial tension test. The rubber that is used in this work is

carbon black filled natural rubber (Rivlin, 1992). The values of the coefficients are:

C10 = 0.03622 and C01 = -0.00335.

All the design variables must satisfy the design range which could be considered

as inequality constraints that dictates the lower and upper bound of these variables. Each

one of these ranges that specify the upper and lower limit of the design variables are

considered as inequality constraints and are incorporated in the finite element optimizer.

The static deflection that is due to the static weight of the engine is measured along the

axis of gravity is given by:

𝐹𝐹𝑔𝑔
𝛿𝛿 = � � (6.4)
𝑘𝑘

In Eq. (6.4), Fg represents the static weight of the engine and k represents the stiffness in

the gravity direction.

The optimization problem described by Eq. (4.8) is solved using the SQP

technique that employs a MATLAB built in function (fmincon) to minimize the value of
 121

the objective function formulated in Eq. (4.7). Once the operation is complete, the design

vector that corresponds to the optimum value of the objective function is known. The

design vector includes the stiffness values of the engine mount. The second part of the

problem is initiated by setting the objective function described in Eq. (6.1) to minimize

the difference between the desired stiffness values obtained from the first optimization

done through the dynamic analysis and the stiffness values obtained from the geometric

shape of the mount. As a first step, a static analysis is performed to determine the

deflections. The objective function employs the mount geometric data as the design

vector. The design vector consists of 𝑡𝑡𝑠𝑠 , 𝑡𝑡𝑧𝑧 , 𝑡𝑡𝑟𝑟 and 𝜃𝜃. The shape optimization takes into

account the range of the design variables that acts like lower and upper bounds. These

bounds are as shown in Eq. (6.5) below.

0.3 ≤ 𝑡𝑡𝑟𝑟 ≤ 0.59

0.3 ≤ 𝑡𝑡𝑠𝑠 ≤ 1.5 (6.5)

0.5 ≤ 𝑡𝑡𝑧𝑧 ≤ 1.77

− 𝜋𝜋�18 ≤ 𝜃𝜃 ≤ − 𝜋𝜋�6

6.4 Examples

In this section two examples dealing with shape optimization of mounts are

presented. The first example is based upon the six DOF model formulated in section

4.1.1. The second example will be based upon the twelve DOF model formulated in

section 4.1.2. For both of these examples, the final geometrical shape of the mount will

be determined. The systems in both of the examples are subjected to different types of

loading conditions.
 122

6.4.1 Example I

The example presented herein is a continuation of the example presented in

section 4.2.2.1. This example is based on the six DOF model presented in section 4.1.1.

The input load vector corresponds to the engine shaking load at the steady engine speed

of 4000 rpm. The design variables resulting from the optimization process are shown in

Table 4.3. The data given in Table 4.3 which represent the stiffness values of the mounts

are used to set up the objective function presents in Eq. (6.1). The optimization problem

is formulated and solved using the finite element software ANSYS®. The design vector

consists of the four parameters that fully describe the geometry of the mount

𝑡𝑡𝑠𝑠 , 𝑡𝑡𝑧𝑧 , 𝑡𝑡𝑟𝑟 and 𝜃𝜃 as shown in Fig. 6.1. The optimized design variables must satisfy the

lower and upper bounds set in Eq. (6.5). The final values of the design vector are shown

in Table 6.1 along with the minimum value of the objective function. The initial shape of

the mount is shown in Figs. 6.7 and 6.8 and the final optimized shape of the mount is

shown in Figs. 6.9 and 6.10. An analysis was also performed for three different engine

speeds corresponding to idling (800 rpm), steady state cruising (3000 rpm) and over

revving situation (5000 rpm). The results for the shaking force vectors calculated at 800

rpm, 3000 rpm and 5000 rpm are shown in Table 5.1. The corresponding design variable

vector 𝑡𝑡𝑠𝑠 , 𝑡𝑡𝑧𝑧 , 𝑡𝑡𝑟𝑟 and 𝜃𝜃 that corresponds to these different shaking force vectors are shown

in Table 6.2 and Fig. 6.11.

 123

Figure 6.7: Isometric View of the Initial Geometry

Figure 6.8: Front View of the Initial Geometry

 124

Figure 6.9: Isometric View of the Optimized Geometry (Example I)

Figure 6.10: Front View of the Optimized Geometry (Example I)

 125

DV Vs. ω
7

4
theta (rad)
DV

3 tr (in)
ts (in)
2 tz (in)

0
0 1000 2000 3000 4000 5000 6000

ω (rpm)

Figure 6.11: Design Variables Vector Vs. Steady State Speed

Table 6.1: Parameter Optimization Results

Target
Initial Optimized
Stiffness
θ (rad) 6.021 5.9052

tr (in) 0.591 0.4259

Design Variables
ts (in) 0.787 0.9387

tz (in) 1.378 1.4016

kx 3411.3 1065.5 1016.6

Stiffness (lb/in) ky 9183.8 1042.7 1016.6

kz 1852.5 7520.6 7503.5

Obj. Function ψ 104371544.93 3364.83

 126

Table 6.2: Design Variables Vector Corresponding to different Steady State Speeds

ω (rpm)

800 3000 5000

θ (rad) 6.00189 5.8888 5.9394

tr (in) 0.57968 0.4110 0.4795

ts (in) 1.18617 0.9951 0.9154

tz (in) 1.46379 1.4299 1.4100

Table 6.3: Transmitted Loads at Different Operating Speeds

ω (rpm)

800 3000 5000

Transmitted Load 188.807 266.812 258.616

Transmitted Load for

Optimum values at 3000 189.727 266.812 259.62
rpm

Figure 6.11 shows the variation in the optimized mount geometry parameters

𝑡𝑡𝑠𝑠 , 𝑡𝑡𝑧𝑧 , 𝑡𝑡𝑟𝑟 and 𝜃𝜃 as a function of the engine speed. Table 6.3 shows the optimum

transmitted force at different engine speeds as a result of having the optimum stiffness

values at 3000 rpm which are obtained from the geometrical optimization. As can be

clearly seen from Table 6.3, that there is minimal change to the transmitted loads at 800

rpm and 5000 rpm when selecting an optimum geometry corresponding to 3000 rpm

6.4.2 Example II

The example presented herein is a continuation of the example presented in

section 4.2.3.1. This example is based on the twelve DOF powertrain-swing-arm model
 127

presented in section 4.1.2. The input load vector in this example is a linear combination

of the engine shaking load at the steady speed of 4000 rpm and road loads due to road

load profile #1 shown in Fig. 5.10. The design variables resulting from the optimization

process are shown in Table 4.8. The data given in Table 4.8, which represent the stiffness

values of the mounts are used to set up the objective function presents in Eq. (6.1). The

optimization problem is formulated and solved using the finite element software

ANSYS®. The design vector consists of the four parameters that fully describe the

geometry of the mount 𝑡𝑡𝑠𝑠 , 𝑡𝑡𝑧𝑧 , 𝑡𝑡𝑟𝑟 and 𝜃𝜃 as shown in Fig. 6.1. The optimized design

variables must satisfy the lower and upper bounds set in Eq. (6.5). The final values of the

design vector are shown in Table 6.4 along with the minimum objective function value.

The initial shape of the mount is shown in Figs. 6.7 and 6.8 and the final optimized shape

of the mount is shown in Figs. 6.12 and 6.13.

Figure 6.12: Isometric View of the Optimized Geometry (Example II)

 128

Figure 6.13: Front View of the Optimized Geometry (Example II)

Table 6.4: Parameter Optimization Results

Target
Initial Optimized
Stiffness
θ (rad) 6.021 5.923

tr (in) 0.591 0.454

Design Variables
ts (in) 0.787 0.965

tz (in) 1.378 1.430

kx 3411.3 2410.40 2405

Stiffness (lb/in) ky 9183.8 2400.10 2405

kz 1852.5 1572.80 1564.5

Obj. Function ψ 47047713.13 122.06

The design of a shear (bush) type mount has been obtained using the geometrical

shape optimization using the parameterization technique. The method was done through

utilizing a nonlinear finite element analysis. Part of the design was done using SQP
 129

method in MATLAB® in order to find the target stiffness values. More details regarding

these results are available in sections 4.2.2.1 and 4.2.3.2. As can be noticed from the

results above, the optimum shape of the mount is acceptable and can be used as the final

shape. It is worth mentioning that this approach is applicable for any type of engine

mounts. The stiffness values that are obtained from the shape optimization are slightly

different than those values obtained from the dynamic analysis; however, the shape

obtained from the parametric optimization is acceptable and can be used in real design

situations.

6.5 Summary

In this chapter, a determination of the geometrical shape of the engine mount is

discussed. The shape optimization is performed using ANSYS®. The process begins by

performing dynamic analysis which is represented in chapter 4 and then the shape

optimization is performed. Two different examples are considered in this chapter. The

first example is based on the six DOF model in which the objective function is

formulated using the shaking force vector as the only input load. The second example is

based on the twelve DOF model in which the objective function is formulated using a

combination of the force vector consisting of both the shaking force and forces

transmitted through the tire patch.

 130

7 Chapter 7 - Conclusion

The mount system modeling work presented in this dissertation can be generally

divided into three main areas. The first area deals with development of mathematical

models of the mount and the mounting system that varies in complexity, in order to

examine their vibration isolation quality. The second area makes use of the developed

theoretical models in order to come up with the design of an optimum passive engine

mounting system such that the forces transmitted to the frame to the system are

minimized. The third area explores finding the optimum engine mount geometrical shape

with desired stiffness and damping characteristics.

This chapter summarizes the work that has been done in this dissertation and

provides an outline for possible future work on this topic.

7.1 Theoretical Modeling

Several engine mounting system were discussed in chapter 3 which the model

complexity varied from a simple Voigt model to more complex Voigt model with Bouc-

Wen element and nonlinear stiffness. In all of the models that have been discussed, the

major concern was to eventually enhance the vibration isolation quality. A close attention

was paid to the problem of mechanical snubbing when the isolated mass undergoes large

displacements.

All the models proposed in this work are generic and are applicable to a wide

range of motorcycle layouts. All the models account for motorcycle systems that exhibit

very rigid frames. The overall motorcycle stiffness is affected by the structural

compliance of some components such as the frame, the swing-arm, the powertrain and
 131

the connecting elements between these components. All of these components are

discussed in detail in this dissertation.

Unlike the automobile vibration isolation problem, the motorcycle vibration

isolation depends heavily on the loading conditions where the engine excitation force is

not the only source of vibration; the motorcycle is affected by the engine excitation force

and the road loads as well. In addition, the overall motorcycle stiffness and the ride

quality is strongly affected by the isolation system. The work presented in this

dissertation discusses both loading conditions in order to come up with the optimum

mounting system configuration. The road loads are calculated using the vertical stiffness

and damping characteristics of the tire patch and the displacement profile of the road

surface. The engine excitation loads are computed from the shaking forces present in a V-

Twin engine.

Two major mounting system models were used in this dissertation. The first

model is formulated by considering the mounting system to be a six degree of freedom

system and the second model is formulated by considering the mounting system to be a

twelve degree of freedom model. In both models, the frame is assumed to infinitely rigid.

In the second model, the powertrain assembly and the swing-arm assembly are connected

via a coupler shaft at a pivot point. The second model allows for the road loads to be in

the problem solution.

7.2 Mounting system optimization

The major role of the mounting system in addition to physically mount the

powertrain to the vehicle frame is to provide vibration isolation. It is important to insure

sufficient clearance between the powertrain and the surrounding components. This is
 132

achieved by imposing the appropriate boundaries when designing the mounting system.

The mounting system should be able to isolate the frame under steady state loading

conditions at the same time limiting the maximum excursion of the powertrain under

transient loading conditions.

In chapter 4, the mount models were developed to determine the optimum

mounting system characteristics by minimizing the load transmitted from the engine to

the frame through the mounting system under multiple loading conditions. This objective

was achieved while satisfying the displacement constraints to limit the maximum

excursion at specific locations on the powertrain. In the design process, the mount

location, orientation and stiffness parameters were used as the design vector. It was found

that the mount orientation is significant and very important in achieving enhanced

vibration isolation. The optimization problem was solved using MATLAB®.

7.3 Shape Optimization

Finding the mounting system characteristics, which was done in chapters 4 and 5

has proven to be very important when it comes to vibration isolation. Finding the

optimum geometrical mount shape has proven to be vital also. The shape optimization of

the engine mount is done with the help of a finite element model that employs a nonlinear

analysis technique. This work is done using ANSYS®. The mount geometry is built in

ANSYS® using the major dimensions that will fully describe the geometry. The finite

element model is generated using a suitable element to accommodate the nonlinear

properties of rubber from which the mount is made off. A Money Rivlin rubber model is

used to describe the fully incompressible hyperelastic material behavior of rubber.

 133

The geometrical parameters of the mount such as the mount diameter, rubber

thickness, etc. are used as the design vector. All the assumptions that are necessary to

find the best loading conditions as well as boundary conditions are discussed in chapter 6.

The loading and boundary conditions are important when it comes to considering

different sources of vibration and physically mounting the powertrain on the frame.

7.4 Future Work

There are some issues that have surfaced in the work discussed in this dissertation

which are related to this work. These issues have not been addressed in this work, but

could be investigated in the future research.

All the models that have been discussed in this work were used to optimize the

engine mount which is a passive mount. Future models could be modified such that

engine mounting system consists of active mounts in addition to or in line of passive

mounts. The mounting system could include a variable stiffness and damping properties

that are under a control law based on accelerations at certain points on the frame.

The focus of this work was on the in-plane stiffness properties. The out-of plane

stiffness properties could be considered by considering an enhanced representation of the

front and rear end of the motorcycle.

All the models presented in chapters 4 and 5 of this dissertation are based on

simple Voigt model that employs simple linear spring stiffness. Future work could

consider more complex model that used a nonlinear stiffness. Mechanical snubbing will

be better represented using such models which in turn will enhance the understanding of

the mounting system vibration isolation.

 134

External loads are transmitted through the tire patch to the engine through the

mounting system. The load calculation is based on the vertical stiffness and damping of

the tire which is affected by the displacement profile of the road. An alternative technique

for load estimation that could be used in the future research is treating the structure as a

load transducer. In this technique, strains are measured by placing strain gauges at some

specific locations on the frame which can be used to provide a history of the loads acting

on it. A finite element analysis can be used to find the appropriate locations and

orientations of the strain gauges to be mounted on the frame.

 135

References

1. Akanda, A. and Adulla, C., 2005, “Engine Mount Tuning for Optimal Idle and
Road Shake Response of Rear-Wheel-Drive Vehicles,” Society of Automotive
Engineers, technical paper series 01-2528.

2. Akanda, A. and Adulla, C., 2006, “Application of Evolutionary Commutation in

Automotive Powertrain Mount Tuning,” Shock and Vibration, Vol. 13, pp. 85-
102.

3. Ali, A., Hosseini, M and Sahari, B. B., 2010, “A Review of Constitutive Models
for Rubber Like Materials,” American Journal of Engineering and Applies
Science, Vol. 3, pp. 232-239.

4. ANSYS 12.0, Help Documentation, 2009, ANSYS Inc.

5. Atkinson, A. C. and Donev, A. N., 1992, Optimum Experimental Design, Oxford

Science Publications.

6. Bretl, J., 1993, “Optimization of Engine Mounting Systems to Minimize Vehicle

Vibration,” SAE, paper no. 931322, pp. 475-482.

7. Carfagni, M., Lenzi, E. and Pierini, M., 1998, “The Loss Factor as a Measure of
Mechanical Damping,” Proceedings of the 16th International Modal Analysis
Conference, pp. 580-584.

8. Chen, C. T., 2001, Digital Design Processing – Spectral Computation and Filter
Design, Oxford University Press, Chapters 2, 3 and 4.

9. Cocco, G., 2001, Motorcycle Design and Technology, Giorgia Nada.

10. Courteille, E. and Mortier, F., 2005, “Multi-Objective Robust Design

Optimization of an Engine Mounting System,” SAE, paper no. 01-2412.

11. Crede, C. E., 1965, Shock and Vibration Concepts in Engineering Design,
Prentice Hall Inc Englewood Cliffs, N.J.

12. Crede, C. E., 1951, Vibration and Shock Isolation, John Wiley and Sons Inc, New
York, 1951.

13. Derby, F. T., 1973, “Decoupling the Three Translational Modes from the Three
Rotational Modes of a Rigid Body Supported by Four Corner-Located Isolators,”
Shock and Vibration Bulletin, Part 4, Vol. 43, pp. 91-108.

14. Dhingra, A. K. and Hunter, T. G., 2003, “Optimum Experimental Design of A

General Purpose Load Transducer,” NAFEMS World Congress, Orlando, FL.
 136

15. Ford, M. D., 1985, “An Analysis and Application of a Decoupled Engine Mount
System for Idle Isolation”, SAE, paper no. 850976, pp. 133-142.

16. Garmaroudi, M. A. and Mosayebi, J., 2008, “Design and Optimization of Engine
Mount,” International Review of Mechanical Engineering, Vol. 2, No. 5, pp. 682-
692.

17. Geck, P. E. and Patton, R. D., 1984, “Front Wheel Drive Engine Mount
Optimization,” Society of Automotive Engineering, No. 840736, pp. 123-134.

18. Harris, C. M. and Crede, C. E., 1961, Shock and Vibration Handbook, McGraw-
Hill New York, Chapter 30, pp. 18-38.

19. Ikhouane, F., Manosa, V. and Rodellar J., 2006, “Dynamic Properties of the
Hysteretic Bouc-Wen Model,” Systems and Control Letters. Vol. 56, pp. 197-205.

20. Iwahara, M., 1999, “The Optimum Layout of Engine Mounting by Dynamic
Analysis,” SAE, paper no. 01-3717.

21. Jeong, T. and Singh, R., 2000, “Analytical Methods of Decoupling the
Automotive Engine Torque Roll Axis,” Journal of Sound and Vibration, Vol.
234, pp. 85-114.

22. Jianqiao, Y. and Williams, F. W., 1997, “Quadratic Representation of Nonlinear

Dynamic Stiffness Matrix and Related Eigenvalue Problems,” Computer Methods
in Applied Mechanics and Engineering, Vol. 146. pp. 313-323.

23. Kaul, S., 2006, Modeling Techniques for Vibration Isolation in Motorcycles, PhD
Thesis, University of Wisconsin, Milwaukee.

24. Kaul, S., Dhingra, A. K., and Hunter, T. G., 2005, “Two Approaches for
Optimum Design of Motorcycle Engine Mount Systems,” Engineering
Optimization, Vol. 37, No. 3, pp. 307-324.

25. Kaul, S., Dhingra, A. K., and Hunter, T. G., 2007, “Frame Flexibility Effects on
Engine Mount Optimization for Vibration Isolation in Motorcycles,” Journal of
Vibration and Acoustics, Vol. 129, pp. 590-600.

26. Kim, J. J. and Kim, H. Y., 1997, “Shape Design of an Engine Mount by a Method
of Parameter Optimization,” Computers and Structures, Vol. 65, No.5, pp. 725-
731.

27. Lee, D. H., Hwang, W. S. and Kim, C. M., 2002, “Design Sensitivity Analysis
and Optimization of an Engine Mount System Using an FRF Based
Substructuring Method,” Journal of Sound and Vibration, Vol. 255, pp. 383-397.
 137

28. Liu, C. Q., 2003, “A Computerized Optimization Method of Engine Mounting

System,” SAE, paper no. 01-1461.

29. Masroor, S. A. and Zahary, L. W., 1990, “Design an All-Purpose Force

Transducer,” Experimental Mechanics, Vol. 31, No.1, pp. 33-35.

30. MATLAB User Guide, Version 7.0, Release 14, Math Works.

31. Norton, R. L., 2004, Design of Machinery: An Introduction to the Synthesis and
Analysis of Mechanisms and Machines, 3rd edition, McGraw Hill.

32. Pacejka, H. B., and Bakker, E., 1993, “The Magic Formula Tyre Model,” Vehicle
System Dynamics, Vol. 21, pp. 1-18.

33. Pacejka, H. B., and Besselink, I. J. M., 1997, “Magic Formula Tyre Model with
Transient Properties,” Vehicle System Dynamics, Vol. 27, pp. 234-249.

34. Pacejka, H. B., 1995, “The Role of Tyre Dynamic Properties,” Smart Vehicle, pp.
55-68.

35. Pacejka, H. B., 2002, Tyre and Vehicle Dynamics, Butterworth-Heineman.

36. Paul, B., 1979, Kinematics and Dynamics of Planner Machinery, Prentice Hall
INC, New Jersey.

37. Rao, S. S., 2000, Engineering Optimization Theory and Practice, 3rd edition, John
Wiley & Sons, Chapter 7.

38. Rao, S. S., 1990, Mechanical Vibrations, Addison Wesley, Reading, MA.

39. Rivlin, R. S., 1992, “The Elasticity of Rubber,” Rubber Chem. Technol. 65, G51-
G66.

40. Scharnhorst, T. and Pain, T. H., 1978, “Finite Element Analysis of Rubber-Like
Materials by Mixed Model,” International Journal for Numerical Methods in
Engineering, Vol. 12, pp. 665-676.

41. Singh, R., 2000, “Dynamic Design of Automotive Systems: Engine Mounts and
Structural Joints,” Sadhana, Vol. 25, Part 3, pp. 319-330.

42. Singh, R. and Royston, T. J., 1996, “Optimization of Passive and Active Non-
linear Vibration Mounting Systems Based on Vibration Power Transmission,”
Journal of Sound and Vibration, Vol. 194, pp. 295-316.
 138

43. Singh, R. and Park, J. Y., 2007, “Effect of Engine Mount Damping on the Torque
Roll Axis Decoupling,” SAE International, No. 01-2418, pp. 1-9.

44. Snyman, J. A. and Heyns, P.S. and Vermenulen, P. J., 1995, “Vibration Isolation
of a Mounted Engine Through Optimization,” Mechanical and Machine Theory,
Vol. 30, No.1, pp. 109-118.

45. Soroka, W. W., 1965, “Hysteretically Damped Vibration Absorber and an

Equivalent Electrical Circuit,” Experimental Mechanics, Vol. 5, pp. 53-58.

46. Spiekermann, C. E., Radcliff, C. J. and Goodman, E. D., 1985, “Optimal Design
and Simulation of Vibrational Isolation Systems,” Journal of Mechanisms,
Transmission and Automation in Design, Vol. 107, pp. 271-276.

47. Suh, M. W., Shim, M. B., Kim, M. S. and Hong, S. K., 3003, “Multidisciplinary
Design of Engine Mounts with Consideration of the Driveline”, Automobile
Engineering, Vol. 217, pp. 107-114.

48. Sui, S. J., 2003, “Powertrain Mounting design Principles to Achieve Optimum
Vibration isolation with dimension Tools,” Society of Automotive Engineers,
paper no. 01-1476.

49. Swanson, S. R., 1985, “Large Deformation Finite Element Calculations for
Slightly Compressible Hyperelastic Materials,” Computers and Structures, Vol.
21, pp. 81-88.

50. Tao, J. S. and Liu, G. R and Lam, K. Y., 2000, “Design Optimization of Marine
Engine-Mount System,” Journal of Sound and Vibration, Vol. 253, No.3, pp. 477-
494.

51. Timpner, F. F., 1965, “Design Consideration in Engine Mounting,” SAE, paper
no. 650093 (966B).

52. Wen-ku, S., Jiang-hua, F., Hao, Y. and Teng, T., 2009, “Multi Objective
Optimization of Powertrain Mounting Based on MATLAB,” Second Asia-Pacific
Conference on Computational Intelligence and Industrial Application, No. 978-
4244-4607, pp. 484-487.

53. Wickham, M. J., Riley, D. R. and Nachtsheim, C. J., 1995, “Integrating Optimal
Experimental Design of a Multi Axis Load Transducer,” Journal of Engineering
for Industry, Vol. 117, pp. 400-405.

54. Ye, M., Wang, X., 2007, “Parameter Identification of the Bouc-Wen Hysteresis
Model Using Particle Swarm Optimization,” Smart Materials and Structures, Vol.
16, pp. 2341-2349.
 139

55. Yunhe, Y., Naganathan, G. N. and Dukkipati V. R., 2001, “A Literature Review
of Automotive Vehicle Engine Mounting System,” Mechanism and Machine
Theory, Vol. 36, pp. 123-142.

56. Zhang J., Richards, M. C., 2006, “Dynamic Analysis and Parameter Identification
of a Single Mass Elastomeric Isolation System Using a Maxwell-Voigt Model,”
Journal of Vibration and Acoustics, Vol. 128, pp. 713-721.
 140

APPENDIX A

The transformation matrix (A) is used extensively throughout this dissertation to

transform the stiffness matrix values from one coordinate to another. This transformation

matrix is a composition of three successive rotations through angles 𝜃𝜃1 , 𝜃𝜃2 , and 𝜃𝜃3 about

the x, y and z axes respectively of a global coordinate system (Crede, 1965). The

transformation matrix is defined as follows:

𝐴𝐴 = 𝑅𝑅𝑧𝑧 (𝜃𝜃3 ) 𝑅𝑅𝑦𝑦 (𝜃𝜃2 ) 𝑅𝑅𝑥𝑥 (𝜃𝜃1 ) (𝐴𝐴. 1)

Substituting all the rotation matrices represented in Eq. (A.1) yields the following:

𝐶𝐶𝜃𝜃3 −𝑆𝑆𝜃𝜃3 0 𝐶𝐶𝜃𝜃2 0 𝑆𝑆𝜃𝜃2 1 0 0

𝐴𝐴 = � 𝑆𝑆𝜃𝜃3 𝐶𝐶𝜃𝜃3 0� � 0 1 0 �� 0 𝐶𝐶𝜃𝜃1 −𝑆𝑆𝜃𝜃1 � (𝐴𝐴. 2)
0 0 1 −𝑆𝑆𝜃𝜃2 0 𝐶𝐶𝜃𝜃2 0 𝑆𝑆𝜃𝜃1 𝐶𝐶𝜃𝜃1

In Eq. (A.2), 𝐶𝐶𝜃𝜃𝑖𝑖 = 𝑐𝑐𝑐𝑐𝑐𝑐𝜃𝜃𝑖𝑖 and 𝑆𝑆𝜃𝜃𝑖𝑖 = 𝑠𝑠𝑠𝑠𝑠𝑠𝜃𝜃𝑖𝑖 . The expression represented in Eq. (A.2) can

be rewritten in the following form:

𝐶𝐶𝜃𝜃2 𝐶𝐶𝜃𝜃3 −𝐶𝐶𝜃𝜃1 𝑆𝑆𝜃𝜃3 + 𝑆𝑆𝜃𝜃1 𝑆𝑆𝜃𝜃2 𝐶𝐶𝜃𝜃3 𝑆𝑆𝜃𝜃1 𝑆𝑆𝜃𝜃3 + 𝐶𝐶𝜃𝜃1 𝑆𝑆𝜃𝜃2 𝐶𝐶𝜃𝜃3
⎡ ⎤
⎢ ⎥
𝐴𝐴 = ⎢ 𝐶𝐶𝜃𝜃2 𝑆𝑆𝜃𝜃3 𝐶𝐶𝜃𝜃1 𝐶𝐶𝜃𝜃3 + 𝑆𝑆𝜃𝜃1 𝑆𝑆𝜃𝜃2 𝑆𝑆𝜃𝜃3 −𝑆𝑆𝜃𝜃1 𝐶𝐶𝜃𝜃3 + 𝐶𝐶𝜃𝜃1 𝑆𝑆𝜃𝜃2 𝑆𝑆𝜃𝜃3 ⎥ (𝐴𝐴. 3)
⎢ ⎥
⎣ −𝑆𝑆𝜃𝜃2 𝑆𝑆𝜃𝜃1 𝐶𝐶𝜃𝜃2 𝐶𝐶𝜃𝜃1 𝐶𝐶𝜃𝜃2 ⎦

The expression given in Eq. (A.3) is the same expression for the transformation matrix

that has been used throughout this dissertation.

Beside the rotation matrix expressed in Eq. (A.3), Euler angles could be used as

an alternative way of computing the rotation matrix (A). Euler angle transformation

matrix is composed by using a rotation through angle φ about the z axis followed by a

rotation through angle θ about the y axis followed by a rotation through angle of ψ about

the z axis. All of the above rotations are performed about the latest frame or the current

frame. The composition is expressed as follows:

 141

𝐴𝐴 = 𝑅𝑅𝑧𝑧 (𝜙𝜙) 𝑅𝑅𝑦𝑦 (𝜃𝜃) 𝑅𝑅𝑧𝑧 (𝜓𝜓) (𝐴𝐴. 4)

Eq. (A.4) can be rewritten as follows:

𝐶𝐶𝐶𝐶 −𝑆𝑆𝑆𝑆 0 𝐶𝐶𝐶𝐶 0 𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶 −𝑆𝑆𝑆𝑆 0

𝐴𝐴 = � 𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶 0� � 0 1 1 � � 𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶 0� (𝐴𝐴. 5)
0 0 1 −𝑆𝑆𝑆𝑆 0 𝐶𝐶𝐶𝐶 0 0 1

After computing the rotation matrices represented in Eq. (A.5), the following

transformation matrix results:

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 −𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶

⎡ ⎤
⎢ ⎥
𝐴𝐴 = ⎢𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 + 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 −𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 + 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 ⎥ (𝐴𝐴. 6)
⎢ ⎥
⎣ −𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶 ⎦

Other representations such as the Bryant angles or roll-pitch-yaw angles can also

be used in order to construct the transformation matrix.

 142

APPENDIX B

A comprehensive tire model has been developed by Pacejka to represent the tire

dynamics. This model which is referred to as the magic formula has been used

extensively in the motorcycle industry. This model generates a set of equations for the

forces and moments as a result of the interaction of the tire with road surface which is

calculated at different slip conditions. These slip conditions contains a pure longitudinal

slip, pure lateral slip or a combination of the mentioned slip conditions.

The governing equation of the Pacejka tire model for a specific vertical load and a

camber angle is as follows:

𝑦𝑦(𝑥𝑥) = 𝐷𝐷 𝑠𝑠𝑠𝑠𝑠𝑠�𝐶𝐶 𝑡𝑡𝑡𝑡𝑡𝑡−1 �𝐵𝐵 𝑥𝑥 − 𝐸𝐸�𝐵𝐵 𝑥𝑥 − 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝐵𝐵 𝑥𝑥)��� (𝐵𝐵. 1)

In Eq. (B.1), y is the output variable which could be either the longitudinal force or the

lateral force or the aligning torque and x is the input variable which could be the slip ratio

or the slip angle. Eq. (B.1) generates a curve that passes through the origin. In order for

this curve to offset with respect to the origin, a shift coordinate system is introduced as

follows:

𝑌𝑌(𝑋𝑋) = 𝑦𝑦(𝑥𝑥) + 𝑆𝑆𝑣𝑣

𝑥𝑥 = 𝑋𝑋 + 𝑆𝑆ℎ (𝐵𝐵. 2)

In Eq. (B.2), 𝑆𝑆𝑣𝑣 and 𝑆𝑆ℎ are the two shift parameters along the x and y axes respectively, Y

represents the output variable which could be the longitudinal force Fx, or the lateral

force Fz or the aligning torque My. X represents the input variables which could as a result

of the lateral or longitudinal slip. Fig. B.1 shows the Pacejka curve with the terms listed

above. B, C, D and E are semi-empirical constants with physical meaning associated to

each constant. For example B, C and D represent the cornering stiffness of the tire. These
 143

constants are function of the wheel load, slip, slip angle and slip ratio. The computation

of these constants is based on the experimental results multiple wheel loads, slip angles,

slip ratios and chamber angles.

−𝑆𝑆ℎ D 𝜋𝜋
𝐷𝐷 sin � 𝐶𝐶�
2
tan−1 (𝐵𝐵𝐵𝐵𝐵𝐵)

−𝑆𝑆𝑣𝑣 𝑥𝑥𝑚𝑚 Slip %

Figure B.1: Pacejka Tire Model Plot [35]

 144

APPENDIX C

Some models discussed in this dissertation are developed by assuming that the

powertrain is connected to the swing-arm via a coupler. This model is presented in

section 3.4.2. The coupler shaft is modeled as a simply supported beam with bearing

supports at both ends of the beam.

The stiffness matrix of the coupler shaft Kc, is developed by using a two-element

finite element model. Each node is assigned six degrees of freedom. The complete

stiffness matrix is 18 x 18. By imposing the boundary conditions which consists of

translational and rotational displacements, the stiffness matrix is reduced to a diagonal

matrix. All the translational degrees of freedom of the end points are restrained for the

coupler shaft. Zero moment inputs are used at the end nodes to reduce the overall

stiffness matrix to six degrees of freedom at the center node in which assigned at the

midpoint of the shaft.

𝐾𝐾𝑐𝑐

24 𝐸𝐸1 𝐼𝐼𝑦𝑦1
⎡ 3 0 0 0 0 0⎤
⎢𝐿𝐿1 (4 + 𝜙𝜙𝑧𝑧1 ) ⎥
⎢ 24 𝐸𝐸1 𝐼𝐼𝑧𝑧1 ⎥
0 0 0 0 0
⎢ 𝐿𝐿31 �4 + 𝜙𝜙𝑦𝑦1 � ⎥
⎢ 2𝐴𝐴1 𝐸𝐸1 ⎥
⎢ 0 0 0 0 0⎥
⎢ 𝐿𝐿1 ⎥
2
= ⎢ �4 + 𝜙𝜙𝑦𝑦1 � ⎥ (𝐶𝐶. 1)
⎢ � 2 � 2𝐸𝐸1 𝐼𝐼𝑧𝑧1 ⎥
⎢ −�2 − 𝜙𝜙𝑦𝑦1 � ⎥
⎢ 0 0 0 0 0⎥
𝐿𝐿1 �1 + 𝜙𝜙𝑦𝑦1 � + �4 + 𝜙𝜙𝑦𝑦1 �
⎢ ⎥
⎢ (4 + 𝜙𝜙𝑧𝑧1 )2 ⎥
� � 2𝐸𝐸 𝐼𝐼
1 𝑦𝑦1
⎢ −(2 − 𝜙𝜙𝑧𝑧1 )2 ⎥
0 0 0 0 0
⎢ 𝐿𝐿1 (1 + 𝜙𝜙𝑧𝑧1 ) + (4 + 𝜙𝜙𝑧𝑧1 ) ⎥
⎣ 0 0 0 0 0 0⎦
 145

In Eq. (C.1), the diagonal terms of Kc are referred to as 𝑘𝑘𝑐𝑐𝑐𝑐 , 𝑘𝑘𝑐𝑐𝑐𝑐 , 𝑘𝑘𝑐𝑐𝑐𝑐 , 𝑘𝑘𝑐𝑐𝑐𝑐𝑐𝑐 , 𝑘𝑘𝑐𝑐𝑐𝑐𝑐𝑐 and 𝑘𝑘𝑐𝑐𝑐𝑐𝑐𝑐

respectively in the section 4.2.3.1. 𝜙𝜙𝑦𝑦1 and 𝜙𝜙𝑧𝑧1 are defined as follows:

12 𝐸𝐸1 𝐼𝐼𝑧𝑧1 𝐴𝐴1 𝑟𝑟𝑧𝑧1 2

𝜙𝜙𝑦𝑦1 = = 24(1 + 𝜈𝜈1 ) � � (𝐶𝐶. 2)
𝐺𝐺2 𝐴𝐴𝑠𝑠𝑠𝑠1 𝐿𝐿21 𝐴𝐴𝑠𝑠𝑠𝑠1 𝐿𝐿1

12 𝐸𝐸1 𝐼𝐼𝑦𝑦1 𝐴𝐴1 𝑟𝑟𝑦𝑦1 2

𝜙𝜙𝑧𝑧1 = = 24(1 + 𝜈𝜈1 ) � � (𝐶𝐶. 3)
𝐺𝐺2 𝐴𝐴𝑠𝑠𝑠𝑠1 𝐿𝐿21 𝐴𝐴𝑠𝑠𝑠𝑠1 𝐿𝐿1

In Eq. (C.2) and Eq. (C.3), 𝑟𝑟𝑧𝑧1 and 𝑟𝑟𝑦𝑦1 are the radii of gyration, 𝜈𝜈1 is the passion ratio and

𝐴𝐴𝑠𝑠𝑠𝑠1 and 𝐴𝐴𝑠𝑠𝑠𝑠1 are the effective cross sectional areas in shear. In the above equations, 𝐸𝐸1

is the modulus of elasticity, 𝐴𝐴1 is the cross sectional area, 𝐺𝐺1 is the shear modulus of the

shaft, 𝐼𝐼𝑦𝑦1 and 𝐼𝐼𝑧𝑧1 are the area moments of inertia in which they are equal since the shaft

cross sectional area is circular and 𝐿𝐿1 is the half the length of the shaft.
 146

APPENDIX D

The engine is modeled as a rigid body of mass M which is attached to the frame

by means of the mounting system as shown in Fig. D.1. The origin of the global

coordinate system G is located at the center of mass of the engine C. However, this not

necessarily true when the engine is idling. The Z-axis of the global coordinate system is

parallel to the crankshaft and the X axis is in the vertical direction.

The general translational EOM of the engine is given by:

𝑀𝑀 𝑟𝑟̈ = 𝑓𝑓 (𝐷𝐷. 1)
𝑇𝑇
In Eq. (D.1), M is the mass of the engine and 𝑓𝑓 = �𝑓𝑓𝑥𝑥 , 𝑓𝑓𝑦𝑦 , 𝑓𝑓𝑧𝑧 � is the sum of all forces

acting on the engine block. 𝑟𝑟 = [𝑥𝑥, 𝑦𝑦, 𝑧𝑧]𝑇𝑇 is the position of the center of mass C.

The general rotation EOM is given by:

𝐼𝐼𝜔𝜔̇ + 𝜔𝜔 × 𝐼𝐼𝐼𝐼 = 𝜂𝜂 (𝐷𝐷. 2)

𝑇𝑇
In Eq. (D.2), 𝜂𝜂 = �𝜂𝜂𝑥𝑥 , 𝜂𝜂𝑦𝑦 , 𝜂𝜂𝑧𝑧 � , is the sum of the moments of the individual forces about

𝑇𝑇
C, 𝜔𝜔 = �𝜃𝜃̇𝑥𝑥 , 𝜃𝜃̇𝑦𝑦 , 𝜃𝜃̇𝑧𝑧 � is the angular velocity with 𝜃𝜃𝑥𝑥 , 𝜃𝜃𝑦𝑦 , 𝑎𝑎𝑎𝑎𝑎𝑎 𝜃𝜃𝑧𝑧 are the rotational angles

about the x, y and z axis respectively. I is the inertia tensor which is given by Eq. (D.3).

𝐼𝐼𝑥𝑥𝑥𝑥 −𝐼𝐼𝑥𝑥𝑥𝑥 −𝐼𝐼𝑥𝑥𝑥𝑥

𝐼𝐼 = �−𝐼𝐼𝑦𝑦𝑦𝑦 𝐼𝐼𝑦𝑦𝑦𝑦 −𝐼𝐼𝑦𝑦𝑦𝑦 � (𝐷𝐷. 3)
−𝐼𝐼𝑧𝑧𝑧𝑧 −𝐼𝐼𝑧𝑧𝑧𝑧 𝐼𝐼𝑧𝑧𝑧𝑧

There are two types of forces acting on the engine body that one needs to be

aware of. The first one is the shaking force and moments that are generated as a result of

the movement of the internal components in the cylinder. The second one is the reaction

force that is applied to the engine at each supporting mount.

 147

Figure D.1: Rigid body engine model

Consider an engine that is supported by Nm mounts located at (𝑥𝑥𝑚𝑚𝑚𝑚 , 𝑦𝑦𝑚𝑚𝑚𝑚 , 𝑧𝑧𝑚𝑚𝑚𝑚 ),

where 𝑖𝑖 = 1, 2, … , 𝑁𝑁𝑚𝑚 . It is assumed the three principal stiffness �𝑘𝑘𝑥𝑥𝑥𝑥 , 𝑘𝑘𝑦𝑦𝑦𝑦 , 𝑘𝑘𝑧𝑧𝑧𝑧 � of each

mount are independent of each other. The force moments that are exerted by the elastic

mounts on the engine is given by Eq. (D.4). and Eq. (D.5). It is assumed that the

displacements at the supports are small compared to their distance from the center of

gravity C.

𝑓𝑓𝑥𝑥𝑥𝑥𝑥𝑥 = −𝑘𝑘𝑥𝑥𝑥𝑥 �𝑥𝑥 + 𝑧𝑧𝑚𝑚𝑚𝑚 𝜃𝜃𝑦𝑦 − 𝑦𝑦𝑚𝑚𝑚𝑚 𝜃𝜃𝑧𝑧 �

𝑓𝑓𝑦𝑦𝑦𝑦𝑦𝑦 = −𝑘𝑘𝑦𝑦𝑦𝑦 (𝑦𝑦 + 𝑥𝑥𝑚𝑚𝑚𝑚 𝜃𝜃𝑧𝑧 − 𝑧𝑧𝑚𝑚𝑚𝑚 𝜃𝜃𝑥𝑥 )

𝑓𝑓𝑧𝑧𝑧𝑧𝑧𝑧 = −𝑘𝑘𝑧𝑧𝑧𝑧 �𝑧𝑧 + 𝑦𝑦𝑚𝑚𝑚𝑚 𝜃𝜃𝑥𝑥 − 𝑥𝑥𝑚𝑚𝑚𝑚 𝜃𝜃𝑦𝑦 �, 𝑓𝑓𝑓𝑓𝑓𝑓 𝑖𝑖 = 1, 2, … , 𝑁𝑁𝑚𝑚 (𝐷𝐷. 4)

𝜂𝜂𝑥𝑥𝑥𝑥𝑥𝑥 = 𝑓𝑓𝑧𝑧𝑧𝑧𝑧𝑧 𝑦𝑦𝑚𝑚𝑚𝑚 − 𝑓𝑓𝑦𝑦𝑦𝑦𝑦𝑦 𝑧𝑧𝑚𝑚𝑚𝑚

𝜂𝜂𝑦𝑦𝑦𝑦𝑦𝑦 = 𝑓𝑓𝑥𝑥𝑥𝑥𝑥𝑥 𝑧𝑧𝑚𝑚𝑚𝑚 − 𝑓𝑓𝑧𝑧𝑧𝑧𝑧𝑧 𝑥𝑥𝑚𝑚𝑚𝑚

𝜂𝜂𝑧𝑧𝑧𝑧𝑧𝑧 = 𝑓𝑓𝑦𝑦𝑦𝑦𝑦𝑦 𝑥𝑥𝑚𝑚𝑚𝑚 − 𝑓𝑓𝑥𝑥𝑥𝑥𝑥𝑥 𝑦𝑦𝑚𝑚𝑚𝑚 , 𝑓𝑓𝑓𝑓𝑓𝑓 𝑖𝑖 = 1,2, … , 𝑁𝑁𝑚𝑚 (𝐷𝐷. 5)

 148

A set of differential equations shown in Eq. (D.6) and Eq. (D.7) may be solved

numerically over a period of time [0, T].

𝑀𝑀𝑥𝑥̈ = 𝑓𝑓𝑥𝑥𝑥𝑥 + 𝑓𝑓𝑥𝑥𝑥𝑥 + 𝑓𝑓𝑥𝑥𝑥𝑥

𝑀𝑀𝑦𝑦̈ = 𝑓𝑓𝑦𝑦𝑦𝑦 + 𝑓𝑓𝑦𝑦𝑦𝑦 + 𝑓𝑓𝑦𝑦𝑦𝑦 (𝐷𝐷. 6)

𝑀𝑀𝑧𝑧̈ = 𝑓𝑓𝑧𝑧𝑧𝑧 + 𝑓𝑓𝑧𝑧𝑧𝑧 + 𝑓𝑓𝑧𝑧𝑧𝑧

𝐼𝐼𝑥𝑥𝑥𝑥 𝜃𝜃̈𝑥𝑥 = �𝐼𝐼𝑦𝑦𝑦𝑦 − 𝐼𝐼𝑧𝑧𝑧𝑧 �𝜃𝜃̇𝑦𝑦 𝜃𝜃̇𝑧𝑧 + 𝜇𝜇𝑥𝑥𝑥𝑥 + 𝜇𝜇𝑥𝑥𝑥𝑥 + 𝜇𝜇𝑥𝑥𝑥𝑥

𝐼𝐼𝑦𝑦𝑦𝑦 𝜃𝜃̈𝑦𝑦 = (𝐼𝐼𝑧𝑧𝑧𝑧 − 𝐼𝐼𝑥𝑥𝑥𝑥 )𝜃𝜃̇𝑧𝑧 𝜃𝜃̇𝑥𝑥 + 𝜇𝜇𝑦𝑦𝑦𝑦 + 𝜇𝜇𝑦𝑦𝑦𝑦 + 𝜇𝜇𝑦𝑦𝑦𝑦 (𝐷𝐷. 7)

𝐼𝐼𝑧𝑧𝑧𝑧 𝜃𝜃̈𝑧𝑧 = �𝐼𝐼𝑥𝑥𝑥𝑥 − 𝐼𝐼𝑦𝑦𝑦𝑦 �𝜃𝜃̇𝑥𝑥 𝜃𝜃̇𝑦𝑦 + 𝜇𝜇𝑧𝑧𝑧𝑧 + 𝜇𝜇𝑧𝑧𝑧𝑧 + 𝜇𝜇𝑧𝑧𝑧𝑧

In Eq. (D.6) and Eq. (D.7), 𝑓𝑓𝑥𝑥𝑥𝑥 , 𝑓𝑓𝑦𝑦𝑦𝑦 , 𝑓𝑓𝑧𝑧𝑧𝑧 , 𝜇𝜇𝑥𝑥𝑥𝑥 , 𝜇𝜇𝑦𝑦𝑦𝑦 and 𝜇𝜇𝑧𝑧𝑧𝑧 are the forces and

moments exerted by the mount on the engine. The terms 𝑓𝑓𝑥𝑥𝑥𝑥 , 𝑓𝑓𝑦𝑦𝑦𝑦 , 𝑓𝑓𝑧𝑧𝑧𝑧 , 𝜇𝜇𝑥𝑥𝑥𝑥 , 𝜇𝜇𝑦𝑦𝑦𝑦 and 𝜇𝜇𝑧𝑧𝑧𝑧

are the inertial forces and moments exerted by the engine. The terms 𝑓𝑓𝑥𝑥𝑥𝑥 , 𝑓𝑓𝑦𝑦𝑦𝑦 , 𝑓𝑓𝑧𝑧𝑧𝑧 , 𝜇𝜇𝑥𝑥𝑥𝑥 ,

𝜇𝜇𝑦𝑦𝑦𝑦 and 𝜇𝜇𝑧𝑧𝑧𝑧 are the forces and moments associated by the balancing masses.

The solution yields dynamic forces represented in Eq. (D.4) and moments

represented in Eq. (D.5) exerted by the mounts on the supporting structure due to the

displacement of the engine supports. A suggested objective function is compiled as the

sum of the squares of the forces over the interval [0, 0.4 sec] as follows:

𝑇𝑇 𝑁𝑁𝑚𝑚
1 2
𝐹𝐹 = � �� �|𝑓𝑓𝑥𝑥𝑥𝑥𝑥𝑥 |2 + �𝑓𝑓𝑦𝑦𝑦𝑦𝑦𝑦 � + |𝑓𝑓𝑧𝑧𝑧𝑧𝑧𝑧 |2 �� 𝑑𝑑𝑑𝑑 (𝐷𝐷. 8)
𝑇𝑇
0 𝑖𝑖=1

The proposed objective function may be used as an alternate function to find the

minimum transmitted internal loads from the engine to the frame in order to achieve the

desired vibration isolation. The transmitted force may be minimized with respect to any

set system parameters such as balancing masses and the associated phase angles.
 149

VITA

Fadi Alkhatib

Place of Birth: Ramallah, Palestine

Education:

Ph.D., University of Wisconsin-Milwaukee, December 2013

Major: Mechanical Engineering
Minor: Engineering Mechanics

M.S., University of Wisconsin-Milwaukee, Dec 2003

Major: Engineering Mechanics

B.S., BirZeit University, Palestine, June 2000

Major: Mechanical Engineering

Publications:

Alkhatib, F., Dhingra, A. K., “Shape Optimization of Engine Mounts for Enhanced
Vibration Isolation,” # IMECE2013-64476, ASME 2013 International Mechanical
Engineering Congress & Exposition.

Alkhatib, F., Dhingra A. K., “Road Loads and Their Effect on the Mount
Characteristics,” Submitted to Journal of Vibration and Control, 2013.

Alkhatib, F., Dhingra A. K., “The Effect of Shaking Forces on the Mount Shape and
Design,” Submitted to International Journal of Advanced Manufacturing and
Technology, 2013.