You are on page 1of 115

A thesis submitted for examination for the degree of Master

of Science in Technology

Espoo 22.05.2020
Supervisor: Professor Luc St-Pierre
Advisors: M.Sc. (Tech) Kleemola Jukka
Aalto University, P.O. BOX 11000, 00076 AALTO
www.aalto.fi
Abstract of master's thesis

Author
Title of thesis

Master programme Code

Thesis supervisor
Thesis advisor(s)
Date Number of pages Language

Abstract

Keywords
Aalto University, P.O. BOX 11000, 00076 AALTO
www.aalto.fi

This thesis is dedicated to the memory of my grandfather, who was a true professional of
nuclear engineering, who inspired me to become an engineer and whose intelligence still
motivate me to pursue my goals.

I would like to express my very great appreciation to Professor Luc St-Pierre for his val-
uable suggestions and useful critiques during the writing of this thesis work. It was a great
privilege and honor to do this thesis and study under his guidance. I would also like to
thank him for his friendship and patience.

I would like to thanks Uusitalo Kari for providing me the opportunity to implement this
thesis work at Moventas Gears OY and smoothing our path to complete this challenging
task. My sincere gratitude is also extended to Kauppinen Petri, Mäkinen Jukka, and Klee-
mola Jukka for inspiring my interest in the development of contact analysis solutions and
their valuable technical support on this project. I also wish to show my sincere apprecia-
tion for the whole R&D team for their consistent help and sharing of knowledge.

Finally, I must express my very profound gratitude to my family for providing me with
unfailing support and continuous encouragement when the times got rough. It was a great
comfort and relief to feel your support and mentoring through all study years.

Artem Dymov
22.05.2020
4

Table of Contents
Abstract
Acknowledgments
Table of Contents ................................................................................................................... 4
Symbols ................................................................................................................................. 6
Abbreviations ......................................................................................................................... 7
1 Introduction .................................................................................................................... 8
1.1 Motivation ............................................................................................................... 8
1.2 Objective ............................................................................................................... 10
1.3 Scope ..................................................................................................................... 10
1.4 Limitations ............................................................................................................ 11
1.5 Commissioning party ............................................................................................ 11
2 Literature review .......................................................................................................... 12
2.1 Gear design fundamentals ..................................................................................... 12
2.1.1 Industrial gearing ........................................................................................... 12
2.1.2 Gear types ...................................................................................................... 13
2.1.3 Gear nomenclature ......................................................................................... 14
2.1.4 Velocity ratio ................................................................................................. 15
2.1.5 Conjugate Action ........................................................................................... 16
2.1.6 Gear involute function ................................................................................... 16
2.1.7 Standardization of gears................................................................................. 17
2.1.8 Path of contact ............................................................................................... 18
2.1.9 Contact ratio ................................................................................................... 20
2.1.10 Forces in Helical Gears .................................................................................. 21
2.2 Transmission Error ................................................................................................ 22
2.2.1 Main features of transmission error spectrum ............................................... 23
2.2.2 Source of transmission error .......................................................................... 24
2.2.3 Evaluation of tooth mesh stiffness ................................................................. 25
2.2.4 Types of transmission error ........................................................................... 25
2.3 Gear noise overview .............................................................................................. 28
2.3.1 Gear rattle ...................................................................................................... 28
2.3.2 Gear whine ..................................................................................................... 29
2.3.3 Overall path to airborne noise ........................................................................ 30
2.3.4 Overview of gear noise research .................................................................... 32
2.4 Methods to reduce gear noise and vibration ......................................................... 33
2.4.1 Gear geometry approach ................................................................................ 33
2.4.2 Micro-geometry of teeth approach ................................................................ 34
2.4.3 Manufacturing quality assurance approach ................................................... 35
2.4.4 Design with load approach ............................................................................ 37
2.5 Challenges in gear design for low noise gear transmissions ................................. 38
2.5.1 Contact stress ................................................................................................. 38
2.5.2 Tooth root stress............................................................................................. 39
2.6 Gear simulation models......................................................................................... 40
2.6.1 Mathematical modeling of the gear system ................................................... 40
2.6.2 Non-linear finite element approximation of the gear system model ............. 42
2.7 Analytical gear teeth contacts analysis with KISSsoft .......................................... 44
2.7.1 The load distribution along the face width .................................................... 45
2.7.2 Loaded tooth contact analysis ........................................................................ 47
5

2.8 Evaluation of affecting factors on analytical solution results ............................... 52


3 Gear tonality evaluation ............................................................................................... 57
3.1 Gear teeth contact analysis using finite element method ...................................... 57
3.1.1 Geometry ....................................................................................................... 57
3.1.2 Material characteristics .................................................................................. 59
3.1.3 Finite element mesh models .......................................................................... 60
3.1.4 Finite element contact .................................................................................... 69
3.1.5 Solver procedure ............................................................................................ 72
3.1.6 Transmission error post-processing ............................................................... 72
3.2 Design of evaluation procedure ............................................................................ 73
3.2.1 Assessment criteria ........................................................................................ 73
3.2.2 Comparison approach .................................................................................... 74
3.2.3 Studied cases .................................................................................................. 75
4 Results .......................................................................................................................... 81
4.1 ........................................... 81
4.1.1 Results of case 1 ............................................................................................ 81
4.1.2 Results of case 2 ............................................................................................ 84
4.1.3 Results of case 3 ............................................................................................ 87
4.1.4 Results of case 4 ............................................................................................ 89
4.2 Stress states ........................................................................................................... 92
4.3 Bearing excitations ................................................................................................ 95
5 Conclusion and recommendations ............................................................................... 97
5.1 Conclusions ........................................................................................................... 97
5.2 Recommendation for future work ......................................................................... 99
Bibliography
Attachments
6

Symbols

speed ratio
gear ratio
number of slices used in modelling of helical gear
number of teeth in pinion and gear
total contact ratio
transverse contact ratio
m module
overlap contact ratio

Lewis form factor


tooth face load factor
dynamic velocity factor
overload factor
size factor
load distribution factor
constant of the pinion offset
shaft parallelism out-of-plane deviation
basic rack factor
correction factor specified in Clause 9 of ISO 6336-1:2019(en)
gear blank factor
[rad/s] rotational speed
[°] helix angle
M [Nm] applied moment
[N] transverse force
[N] nominal transverse force
[N] axial force
[N] tangential load
[N] load at a specific point i
[mm] base radius of pinion and gear
[mm] tip radius of pinion and gear
[m] pitch diameter or reference diameter
[ m] transmission error
[°] rotation angle of pinion and gear
[ ] transmitted power
[m/s] pitch line speed
[ m] peak to peak static transmission error
[ m] crowning height
L [dB(A)] overall noise pressure level
[MPa] maximum surface pressure
[MPa] contact stress
[MPa] bending stress
[°] transverse working pressure angle
[mm] common length of cylinders in contact
[mm] rim thickness
7

[GPa] Young's modulus


[mm] width of the gear tooth or Hertzian half width
[N/ m] mesh stiffness per unit face width or secant contact stiffness
[N/ m] tangent contact stiffness
[N/ m] bending tooth stiffness
[N/ m] tilting tooth stiffness
[N/ m] Hertzian flattening stiffness
[ m] gear tooth deformation under bending
[ m] gear tooth deformation under tilting
[ m] Hertzian flattening
[ m] effective equivalent misalignment
[ m] initial equivalent misalignment
[N/mm] specific load per unit face width
[ m] (equivalent misalignment)
[ m] tolerance on helix slope deviation
[ m] tolerance on helix slope deviation for ISO tolerance class 5
[ m] mesh misalignment due to manufacturing deviations
[ m] equivalent misalignment due to deformations of shafts
[ m] n
[ m] equivalent misalignment due to bearing deformation
[ m] equivalent misalignment due to case deformation
[N/mm* m] pitch line speed
[N/mm* m] theoretical single stiffness
[mm* m/N] minimum value for the flexibility of a pair of meshing teeth

Abbreviations

DIN German Institute for Standardization


IEC International Electrotechnical Commission
ISO International Organization for Standardization
BEM Boundary Element Method
FEM Finite Element Method
R&D Research and development
FFT Fast Fourier Transform
RMS Root Mean Square
TE Transmission Error
STE Static Transmission Error
PPTE Peak to Peak Transmission Error
SAP Start of Active Profile
LPSTC Lowest Point of a Single Tooth Contact
PP Pitch Contact
HPSTC Highest Point of a Single Tooth Contact
EAP End of Active Profile
TCA Tooth Contact Analysis
STEP Standard for Exchange of Product Data
8

1
1.1 Motivation

Nowadays, wind power generation is an alternative to nuclear and coal-fired power plants.
The turbine manufacturers are increasingly producing gearboxes based on lightweight con-
struction, new materials, coatings, and optimized geometries of gear teeth to improve
lifespan and reduce exploitation costs. According to IEC61400-4 (Design requirements for
wind turbine gearboxes [IEC12]), the gearbox must respect acoustic conditions. A modern
design solution should assume that gearboxes must operate quietly and without tonality.

The noise problem is a common issue in power transmission systems. Gear drive is a critical
component of such systems that transfer power by tooth mesh. Gear noise is a quality issue
that causes distraction and annoyance. Excitation caused by gear meshing is the source of
whine or rattle noise when teeth come into contact. Gears should transfer motion smoothly.
However, there are inevitable fluctuations in the real industry gears. These fluctuations occur
every time a tooth passes through the mesh. These fluctuations can be explained and pre-
dicted with transmission error.

Transmission error is an essential mechanism for understanding gearbox sound and vibration
generation that cause mainly by tooth deflection under loading and affected by misalignment
and profile imperfections due to manufacturing. The tooth surface deforms under contact
during a mesh cycle, and the shape deviates from the perfect involute form. There is always
variation in a mesh stiffness that depends on where the load is applied. The tooth is stiffer
near the root and less close to the tip. Change in a tooth mesh stiffness causes the variation
in a gear deflection. Also, there are manufacturing errors in the desired tooth shape. Finally,
gearboxes have a flexible shaft, housing, and bearing parts, that affect gear misalignment.
All these parameters form a transmission error between the actual rotational displacement of
the gears and the expected output of ideal components. As a result, there is a relative dis-
placement between the mating gears. The force between the teeth is generated and leads to
vibration movements through the structure. A high level of mechanical noise is a sign of
weak gearbox and structure design.

Nowadays, the wind energy industry assigns advanced researches and experiments to under-
stand the mechanisms behind the generation of noise emission in mechanical components of
wind turbines. The gearbox of wind turbines is a crucial component to convert and transfer
the rotational power of the rotor to the electric generator with the highest possible efficiency.
The legislation of regions defines the permissible noise emission level that comes from wind
turbines at a specific location. The angular speed fluctuation due to mesh stiffness variation
leads to the torsional acceleration amplification that causes bearing to excite the vibration.
Noise from gearbox transfer to the thin wall structure and resonance amplification of emitted
sound may occur. The main challenge of the industry is to achieve minimum transmission
error and soundproofing with maximum possible power density generated under wind power
loading.
9

There are numerical and analytical solutions on gear contact analysis available with different
assumptions and limitations. The most popular assumptions are the constant load acting on
the gears and the exclusion of varying mesh stiffness. However, the science community ac-
cepts the obtained results of the most numerical and analytical solutions as quantitatively
correct. The validation experiments may give different results because it is complicated to
model the complex non-linear dynamics of gears with all manufacturing errors, surface
roughness, and loading assumptions to solve the equations of motion to extract a meaningful
result.

The primary purpose of contact mechanics simulations is to predict and to reduce TE by


optimizing the tooth profile under a certain assumption. J. Derek Smith [1] shows an analyt-
ical approach to transmission error estimation and study the influence of different profile
modifications on gear excitation. Blankenship et al. [2] performed work on the practical
investigation of a single gear pair dynamic response to different geometrical parameters.
Nowadays, engineers can select a realistic simulation method to study the dynamic response
of transmission systems under different operational conditions and introduce modifications
to a tooth shape.

KISSsoft AG developed a commercial code based on ISO6336-1 and Weber und Banaschek
[3] method to perform contact analysis that used by industry companies for the development
of gear drive systems. An analytical approach has emerged as a quick tool for gear contact
assessment. Still, it ignores gear body geometry, rim thickness, cross-influence, performed
under particular assumption covered in this thesis work, and does not model helical gears
directly. It is essential to understand the efficiency of the analytical method, assess draw-
backs caused by mentioned above restrictions and define correlations and distinctions with
the alternative finite element-based approach that requires a much higher computational cost.
10

1.2 Objective

Gear noise issues have become increasingly relevant to the wind industry. Industry aims to
minimize tonality risks and predict the surrounding pressure level at the early phase of the
design process with advance and accurate simulation tools. It allows satisfying restrictions
of local authorities' regulations and increases wind power density in urbanized areas.

The thesis work aims to identify a range of applicability of the analytical method proposed
by Weber and Banaschek [3] and implemented in KISSsoft AG software in comparison with
the finite element method and recommend virtual testing of transmission error for conceptual
wind gearboxes design use. The analysis of outcomes is aggregated under assessment crite-
ria, developed in this thesis work and based on literature review, for evaluation of the ap-
plicability of contact analysis.

According to Yilmaz [4] and KISSsoft user manual [5], the analytical approach provides
accurate modeling of gear deformation mechanics. Still, it ignores gear body geometry, gear
rim thickness, and cross-influence of power train components. The finite element method
can account for all those effects. The finite element analysis is relatively slow and requires
high computational cost. Hence considering these factors, there is reasonable scope in accu-
racy and robustness assessment of the analytical solution in comparison with the finite ele-
ment approach. It is essential to understand the cause of transmission error in wind gearboxes
to conclude how restrictions of the analytical method may influence on the correlation be-
tween simulated tonality behavior and noise. The central aspect of the thesis study is the
comparative analysis of components affects the resultant transmission error evaluated using
both approaches. Hence correlated results from both methods allow us to analyze the cause
and effect of gearbox components on transmission error curve and generated noise. The work
performed in this thesis is discussed below in detail.

1.3 Scope

This work is mainly focusing on the development of finite element modeling practices in
tonality assessment to find correlations and distinctions with analytical approach provided
by KISSsoft AG software. This thesis work concentrated on the single parallel helical gear
stage described in Section 1.5 and illustrated in figure 1.2. Helical gear allows relatively low
computational cost in comparison with planetary stages and reliability in terms of straight-
forward assessment of single contact patterns instead of multiplies patterns in the planetary
gear system. Literature review of gear design principles and transmission error theory related
to whine noise phenomenon aims to define the optimum theoretical approach to the problem
and references to explain obtained results. The topic of gear noise reduction in terms of ge-
ometry, microgeometry, and loading is carefully overview and discussed to understand the
main affecting factors and their contribution to vibration profile.

As one of the outputs of this thesis works, the finite element method procedure for the de-
termination of transmission error, mesh stiffness, and noise level caused by the gear contact
mechanics in wind turbine gearbox is established and compared with the analytical approach
results. Analytical method according to ISO6336-1 Chapter 7 Annex E is analyzed and per-
formed using KISSsoft software. To identify the class of problems for which analytical
method provides correlated data, study cases based on different product development stages
are developed in Chapter 3 to perform comparative analysis using both ways. Finally, the
11

tendency and difference of transmission error profiles from both methods and the corre-
sponding effect on gear whine noise power are analyzed concerning influencing from gear-
box components and design parameters.

The rest of the thesis is divided into five chapters. Chapter 2 reviews the literature on the
various theoretical aspects, methodology tools, and concepts accounted for in this thesis
work. Chapter 3 outlines the analytical method features and describe the developed finite
element-based contact analysis for the examined gear drive system and relative product com-
ponents. Chapter 4 provides the results of the thesis, including the comparison of both meth-
ods and describes factors that influence on formed vibration profile, corresponding spectrum,
and contact stress states in wind turbine gearbox. Chapter 5 concludes the thesis work by
discussing the accuracy and correlation characteristics of both methods for constant noise
and vibration assessment of complex geared systems.

1.4 Limitations

The following limitations were considered in this thesis work:

Damping effects in the contact analysis and dynamic behavior of bearing, housing
components were not considered. It allows a reduction of the complicity of the mod-
eling process but decreases the accuracy of the obtained results.
The helical stage of the transmission chain was only investigated. The influence of
planetary stages components of gearbox was neglected to reduce computational
costs. The assessed gear drive system consists of a pair of helical gear, shafts, bear-
ing, and housing elements.
Experimental validation of the accuracy of the contact analysis solution was not to
be considered.

1.5 Commissioning party

This master thesis was performed in close collaboration with the conceptual design team of
the R&D department in Moventas Gears OY, located in Jyväskylä, Finland. Moventas is a
leading manufacturer of outstanding gearboxes solutions for sustainable wind energy pro-
duction. The engineering team has in-depth knowledge of drivetrain behavior and continu-
ously work on improvements to increase the competitiveness of wind energy.

As noise levels are an essential concern in the wind power industry, Moventas has already
performed successful experiments to reduce the noise level of its gearboxes by improving
tooth design and manufacturing accuracy. To eliminate any harmful sounds and vibrations
from gearbox components, Moventas specified the thesis work that is a part of a long-term
development task to minimize the tonality risk in the early phase of the design process.
Moventas provided the specifications and design data of elements used in this thesis work
and described in the next section.
12

2
This chapter explains relevant theories of the gear noise and provides an overview of the
research work carried out in the field of gear noise and vibration. This chapter aims to review
the basics of relevant theories, to introduce the results of primary researches on gear noise
problems, and to describe the concept of the transmission error and mesh stiffness with a
strong connection to the geometry of involute gears. The chapter includes a description of
gear design fundamentals, transmission error theory, gear noise aspects, and gear perfor-
mance modeling techniques. Section 2.8 discusses the observed influence on provided re-
sults from simplifications and assumptions implemented in KISSsoft to make the analytical
mathematical model more amenable for calculation.

2.1 Gear design fundamentals


2.1.1 Industrial gearing

The primary purpose of the gear is to vary torque and rotational velocity to transmit power
from one part of the chain to another.

The design should satisfy requirements of cost, weight, noise generation, power output to
correspond to modern legislations and engineering requirements. As a result of researches
and product development in the gear field, there is practical gear-design information availa-
ble in forms of standards, recommendations, and templates. Nowadays, power transmission
technology is continuously evolving and developing virtual prototyping, condition monitor-
ing tools to withstand modern industry requirements.

The selection of gear type depends on design requirements, machine tools available, cost,
quality, and tradition. Gear industry is highly standardized and tends to avoid any abrupt
changes, that comes from new technological development. It is possible to correlate gear
design traditions and standards according to application. The starting point in gear design
selection is application definition. It defines the life requirement, speed of rotation, dimen-
sioning, noise pollution, the accuracy of transmission, power transmission capacity, materi-
als, manufacturing method, and other essential geometry parameters. The geometrical ar-
rangement of the gearbox is the dominant factor in gear type selection.

This thesis work concentrates on helical gears implemented in the drive chain of wind power
industry gearbox. The helical gears are the ones to be selected for the design of parallel axes
transmission at the helical stage of Moventas gearbox.
13

2.1.2 Gear types

This section introduces the main gear types examined in this thesis work.

Spur gears illustrated in figure 2.1 are cylindrical gears that have teeth oriented parallel to
the axis of rotation. They are used to transmit motion from one shaft to another. The tooth
profile usually represents the shape of the involute. Spur gears are accessible in a wide range
of applications due to relatively inexpensive manufacturing and simple design requirements.
Spur gears generate only radial forces and used mainly as low-speed drivers, and noise level
generated by chain is not a concern. Speeds always keep tooth contact with a maximum of
two teeth pairs leading to high variation in tooth stiffness that makes them prone to upper
tooth stress and noise generation.

Figure 2.1 Spur gear pair

Helical gears are involved in power transmission in parallel axes drivers and illustrated in
figure 2.2. It is the usual selection option when the application requires high loading capa-
bilities. Helical gear design is similar to spur gear shape, but the helix angle defines the
incline of teeth. There is the same helix angle magnitude but opposite in orientation in gear
pair design. Correct implementation of helix angle into gear design leads to the gradual
meshing of gear teeth. This effects on loading fluctuation and results in minimal shifts in
tooth stiffness during operation. It makes them generally quieter than spur gears. The main
disadvantage of helical gears is the tendency to high radial and thrust loading on supporting
bearings.

Figure 2.2 Helical gear pair


14

2.1.3 Gear nomenclature

Figure 2.3 Spur gear terminology. Reproduced from [6]

Figure 2.4 Helical gear terminology. Reproduced from [6]


15

The common terminology used in gears today is presented in figure 2.3 and 2.4. The defi-
nition of pitch circle diameter term is the starting point of the most calculations in gear de-
sign. The circle diameter should be tangential for mating pair, so the central distance is the
sum of both gear radius. The pinion is the gear characterized by a smaller number of teeth
in comparison with a gear wheel. The circular pitch is the distance between a correspond-
ing point on neighbors' teeth. Tooth width space is estimated from the circular teeth.

There usually is backlash in the gear drive system. It is measured on pitch circle and defined
as length by which tooth width space differs from the thickness of the engaging tooth. The
addendum is the radial distance between the top surface of the tooth and the pitch circle line
of the gear. The dedendum is known as the radial distance between the pitch circle line and
the bottom surface of the tooth. The involute curve of the gear teeth starts from the base
circle radius. These parameters are relevant prerequisites in the design of tooth contact pro-
files.

2.1.4 Velocity ratio

Figure 2.5 Left side depicted gear geometry in mesh, meanwhile right side shows that contact takes place on
a line tangent to the base circles

The speed ratio is an essential term for understanding gear drive principle of constant angular
speed transition from driving member to driven member of a gear train. If the teeth stay in
contact and normal line passes through the pitch point throughout the gear mesh as illustrated
in figure 2.5, the drive chain save constant rotational velocity ratio in normal to pitch point
direction. This observation is the fundamental law of gear, where equation 2.1 defines gear
ratio. The gear ratio is expressed in terms of teeth number ratio, base circle radius ratio, or
angular velocity ratio.

(2.1)
16

2.1.5 Conjugate Action

When two gears are in mesh, the mating teeth of gears acting on each other to produce rota-
tion, gear teeth pair must generate a constant angular velocity ratio.

Figure 2.6 Conjugate action diagram. Reproduced from [7]

When tooth A acts on driven tooth B, the contact occurs on the point where two surfaces are
unrelated to each other, as illustrated in figure 2.6. All reaction forces are directed along line
a-b, that is always a normal line between acting teeth.

The line a-b is also known as "line of action" and represents the direction of forces acting on
the curved surface. The line goes through pitch point, so the angular velocity ratio between
gear teeth is proportional to the rate of their radii to the pitch point.

The pitch point must stay fixed, and all reaction forces at any time point must pass through
the pitch point to maintain a smooth gear operation.

2.1.6 Gear involute function

The mathematician Leonhard Euler proposed the involute curve profile to form flanks' shape
of a gear tooth. The involute curve is produced by the tracing of a point on the end of a taut
string unwound from the base circle illustrated in figure 2.7. [8]
17

Figure 2.7 Schematic representation of the involute function. Reproduced from [7]

Involute shape fully satisfies the fundamental law of gearing. The cycloid form is an alter-
native to involute profile, which has one disadvantage since the method allows the direction
of force to vary with time because it is sensitive to center distance. Higher noise level and
vibration are generated in gear transmission when force reaction magnitude varies due to
change of application point and corresponding direction of force at different center distances.
The involute profile always acts in the same direction towards pitch point.

It is the type of gear profile that corresponds to constant direction, force and gives low vi-
bration operation. It is a favorite option choose in the gear industry due to relatively simple
manufacturing and its adaptability to a specific modification.

2.1.7 Standardization of gears

The identification of gear geometry solutions is based on their pitch distances. Thus module
is used to define the ratio between the center distance and number of teeth. The module
is millimeters of pitch circle diameter per tooth, as given in equation 2.2:

(2.2)

It is a size dimension, which is used as a reference to calculate other gear design parame-
ters. The angle formed by the normal line of tooth face and the reference circle tangent
known as pressure angle, as depicted in figure 2.8 [9]. Common pressure angles are
The addendum is generally equal to the module. The dedendum is differing
about 15%-25% more than the module of the gear. [10]

Only one flank of gear tooth participates in contact, but to reduce manufacturing cost, im-
prove stability and durability, the gear profile design is usually symmetrical. Geometry pa-
rameter calculations are specified in international standards ISO 6336 and ISO 21771.
They consider the transmittable power, strength calculation, and achievable service life to
reach optimal gear profile design.
18

2.1.8 Path of contact

A spur gear contact happens on a straight line that progresses from the root to the tip of
driving gear and the tip to the root on the driven gear. The straight line is known as a pressure
line. Position 1 corresponds to the first contact position. It is the start of the active profile
(SAP) illustrated in figure 2.8.

Figure 2.8 Contact path a) start of active profile b) lowest point of single tooth contact and end of active profile
at tooth O c) the highest point of single tooth contact
19

Position 2 is the point where tooth O is no longer be the only tooth in contact. It is the highest
point of a single tooth in contact (HPSTC). The position on tooth O in figure 2.8 (b) corre-
sponds to the place where tooth leaves contact and named as the end of active profile (EAP).
When the tooth leaves the contact, and another tooth becomes the only tooth pair in contact,
this position is known as the lowest point of a single tooth in contact. The position depicted
in figure 2.8 (c) shows where single tooth pair contact occurs and where the tangent to pair
base circles intersect the center construction line. It is the pitch point (PP). The length of the
contact path is the distance from the start to the end of the active profile illustrated in figure
2.9. [11]

Figure 2.9 Path of contact on the spur tooth surface

Helical gear contact is a diagonal that starts from the corner of the root to the tip on the
driving gear. This line moves across face width and forms a contact zone illustrated in figure
2.10. The number of teeth in contact and position of contact lines can be defined by axial
and transverse contact ratios, as explained further.

Figure 2.10 Helical diagonal path of contact


20

2.1.9 Contact ratio

When two gears are in mesh, the intersecting addendum lines of both gears restrict the con-
tact zone. As mentioned in Chapter 2.8, the initial contact occurs at SAP and ends at EAP.
The radial distance on the pitch circle between those two points is called the arc of action.
When the circular pitch is smaller than the arc of action, then there is more than one tooth in
a contact. It means that as one tooth is ending being in contact at EAP, another tooth was
already in contact for a specified period. Gear and pinion rotate through the meshing cycle
and act in the identical periodic arc of action for each tooth. The contact ratio provides the
average number of teeth in contact on an arc of action and, in case of spur gear defined as
the ratio of the arc of action to circular pitch.

The estimation of the helical gear contact ratio on the transverse plane is similar to the spur
gear pair case. The helical orientation of teeth faces geometry adds an extra amount of con-
tact ratio. An advance amount is an overlap contact ratio along the length of teeth covering
the width of teeth. The transverse contact ratio for helical gear is the same as in the case of
the spur gear. The total contact ratio is a sum of both contact ratios and given in equation
2.3:

where is a tip radius and is a base radius, is the center distance, is the working
pressure angle, and is the helix angle. For example, the total contact ratio for Moventas
Gears OY helical stage is about 3.44. There at least three gears in contact for 56% of the
total time and four teeth for 44% of the total time.

The contact ratio is usually non-constant due to manufacturing errors in the assembly and
production of the gear pairs. Drivers with low contact ratios are prone to incorrect motion
transfer because of interference and damage due to teeth impact contact. Incorrect motion
leads to an increased level of excitation amplitudes and corresponding noise levels. High-
contact-ratio gears are suitable for normalizing amplitudes at long-life applications. But
these gears are prone to the high loading influence of bending stress at the addendum region.
21

2.1.10 Forces in Helical Gears

The pressure is acting on a tooth surface during motion. Resultant force acts in normal to
tooth surface direction. The influence of friction forces due to teeth surface contact move-
ment is negligible. At helical gears, the resultant force is subdivided in axial force component
and a transversal force component , which magnitude depends on the helix angle ( )
as denoted in equation 2.4 and 2.5:

(2.4)
(2.5)

Transverse force the function is to transfer the torque loading from driving gear to the driven
gear. An axial force is resultant from the torsional deflection of the gear tooth profile. The
transverse force is given in equation 2.6:

(2.6)

The axial force is also denoted in terms of input torque and given equation 2.7. The resulting
output torque is given equation 2.8 and estimated by implementing the fundamental law of
gear:

(2.7)
(2.8)
22

2.2 Transmission Error

Transmission error theory has emerged as an attractive concept for any type of gear profile
under load and manufacturing allowances to obtain the correlation between gear whine, vi-
bration, and deviation from the actual transmission. The excitation forces of gear system
depend on the tooth meshing errors that consist of manufacturing errors and the bending
deformation of the teeth under load. Meshing errors are equivalent to the term of transmis-
sion error. Munro [12] have defined the transmission error as

"The deviation in position of the driven gear (for any given position of the driving gear),
relative to the position that the driven gear would occupy if both gears were geometrically
perfect and undeformed."

Mathematically it is given in equation 2.9:

where is the driving angular gear rotation and corresponds to the driven angular rota-
tion, is tooth numbers.

Transmission error can be reported as angular rotation in units of radians and as linear dis-
placement at the base or pitch radius of driven gear with units of . Pitch line displacement
measurements describe the tangential deformation at the pitch radius, and base circle dis-
placement data provide the deformation along the line of action and the pressure line.
KISSsoft uses a base circle to estimate transmission error data. According to KISSsoft man-
uals [4], transmission error estimated as the amount of the contact point deviation from the
path of contact in .

The linear displacement is useful to correlate results with manufacturing errors that have the
same units and domain of measurements. The linear transmission error value on the driving
wheel can be estimated as denoted in equation 2.10:

where is the base radius of the driving gear. The negative sign of transmission error refers
to gear position in action where driven gear has lagged from the theoretical position of per-
fect conjugate action.

In general, transmission error is a measure of the deviation from the perfect transmission of
exactly uniform angular motion that represents a continuous function of gear rotational po-
sition in the time domain or a frequency domain. Transmission error does not produce gear
whine but considered as its source. The noise generation and its path are discussed in Section
2.3.
23

2.2.1 Main features of transmission error spectrum

The gear vibration spectrum is usually obtained under constant rotational speed and load.
Transmission error is decomposed into the frequency domain by implementing Fast Fourier
Transform (FFT) to assess frequency and harmonics data content and identify problem areas
in the gear train motion, as depicted in figure 2.12. The pair of gears mesh with the same
tooth meshing period and tooth-meshing fundamental frequencies. The gear has N teeth and
rotates with period P, so the tooth-meshing fundamental period is . Tooth meshing fre-
quency is and known as fundamental harmonics frequency. Gear frequency is , so each
Nth rotational harmonic frequency of the gear with N teeth is the tooth-meshing fundamental
frequency. The example of tooth-meshing fundamental frequencies with sidebands and its
amplitudes illustrated in figure 2.11. Sidebands are generated from an amplitude or fre-
quency modulation process. The resultant sideband may also be caused by torque and speed
variation and can be detected on a variable frequency range from tooth harmonics. Mark
defines the source of the tooth-meshing harmonics: [13]

"The geometric deviation of unloaded working-surface from each of the two meshing gears
combined with the elastic deformations of these average working-surfaces is the source of
the tooth-meshing harmonics of a meshing-gear-pair."

Figure 2.11 Example of Spectrum of gear system transmission error. Reproduced from [14]

Manufacturing errors are usually filtered out and analyzed from the measured result. Tooth
errors are estimated via checking tooth frequency and its harmonics. Tooth gives a consistent
response to the rest of the frequency domain. The time-domain is suitable to analyze the
vibration profile formed by different transmission error contributions and to define extreme
points of a function to get a peak to peak values. Houser et al. [15] compared and obtained
a precise correlation between sound power measurements and the summary of harmonics
and sidebands of transmission error in spur and helical gears. It is vital information to modify
gear performance parameters to reduce generated harmonics and corresponding noise.
24

2.2.2 Source of transmission error

Geometry, deflection, and dynamics of the gear train are primary sources of transmission
error. Parametric excitations occur due to variation in mesh stiffness defined by those pa-
rameters.

First, geometry errors consist of manufacturing and assembly imperfections like run-out er-
ror, tooth spacing, profile, lead errors, misalignment of the shaft, and gears. Secondly, the
teeth stiffness varies as gear teeth change their position through the mesh, which generates
corresponding time-varying deflection of teeth in contact. There are always a different num-
ber of teeth in contact. The combined gear mesh stiffness varies as teeth rotate and change
its meshing position. When individual teeth pair lose contact or number of teeth in contact
changes, the combined mesh stiffness varies, and the corresponding change in tooth bending
occurs. It is possible to conclude, that the angular fluctuation from the ideal rotation of a
gear body is caused by variation of combined torsional mesh stiffness and corresponding
teeth deflection through the mesh cycle. The combined mesh stiffness is known as the ratio
of torsional load to the angular rotation of the gear body.

Finally, Gregory et al. [16] report that transmission error consists of steady and varying
components. They validated that a steady part influence only on meshing frequency and its
harmonics, and different components define other parts of the frequency spectrum. This ap-
proach [16] has emerged as an effective solution to conclude that perfect involute gears al-
ways generate periodic transmission error as gear chain moves and mesh stiffness change at
each coordinate.

In simple words, mesh stiffness is the primary source of transmission error. Mesh stiffness
varies due to geometrical imperfections on the teeth surface and changing number of teeth
in contact, that cause acceleration or deacceleration of gears. Corresponding transmission
error represents a continuous variable function with a period equal to one mesh cycle. Design
engineers must minimize the variation in mesh stiffness to reduce generated noise.
25

2.2.3 Evaluation of tooth mesh stiffness

The accuracy of the mesh stiffness estimation is a significant issue in the assessment of the
transmission error vibration profile. Tooth mesh stiffness is usually determined using com-
putational models discussed in this thesis work. It is possible to obtain experimental data of
mesh stiffness by utilizing high precision encoders for detecting angular gear deflections of
mating teeth and validate computational models.

Mesh stiffness determination approaches has been the focus of considerable theoretical [17-
18], numerical [19-20], and experimental work [21]. All studies concentrate on two methods
for approximation of the non-linear force-deflection relationship curve of the gears in mesh.
Cooley et al. [22] stated that the force-deflection curve behaves non-linearly due to elastic
contact mechanics, which vary most dramatically at the point of the number of teeth in con-
tact changes. The first method to determine mesh stiffness is known as secant stiffness or
average slope approach that estimated as ratio total mesh force to total mesh deflection de-
fined by transmission error in equation 2.11. It makes this approach valid for finite element
software that provides total mesh deflection values.

The second method is known as tangent stiffness or local slope approach, that is the first-
order finite difference approximation of non-linear mesh force versus mesh deflection at
certain nominal deflection point. Cooley et al. [22] observed amplitude and shape difference
between both method results for a wide range of used geometries and loading conditions,
and experimentally conclude that average slope method provides accurate results for the
cases where high contact ratio gears, profile modification, bearing forces, loading sharing,
and static windup conditions were considered. The local slope approach is not an accurate
method for mesh stiffness prediction.

The accuracy of the provided results of mesh stiffness is essential for understanding the
sources of vibration profile waveform, that define the accuracy of spectrum prediction and
corresponds generated noise power.

2.2.4 Types of transmission error

The transmission error sources have been the focus of intensive theoretical, numerical, and
experimental work. As a result, the total transmission error is described and assessed in terms
of static, dynamic, kinematic, and manufacturing transmission errors. The studies of this
thesis work are limited to static transmission error due to its main contribution in gear whine
discussed in section 2.4.
26

2.2.4.1 Static transmission error

The gear teeth can be represented as a beam that deflects elastically along with the shaft,
bearing, and housing under loading. Static transmission error considers stiffness of all com-
ponents in the system and study deflection of gears under loading. Teeth deform in the form
of bending at contact and deviates from true involute depending on mesh stiffness variation.
Mark [23] has expressed the stiffness as an area integral over contact zone that includes all
modifications and involute profile form. Static Transmission error studies the mesh stiffness
variation under loading caused by profile parameters, the path of properties, and gear contact
ratio. It studies the ideal profile and does not assume any manufacturing imperfections. Mod-
eling of the gear transmission system to calculate static transmission error receive consider-
able attention to predict peak to peak values on the transmission error curve and the corre-
sponding generated noise by implementing different profile modifications. The typical Static
transmission error has a stepped periodic form. The measured example of static transmission
error of a helical gear pair is illustrated in figure 2.12.

Figure 2.12 The Static Transmission error of ideal helical gear. Reproduced from [24]

2.2.4.2 Dynamic transmission error

The dynamic transmission error estimates gear behavior where significant variation in the
actual length of the contact line occurs due to inertia forces, flank shape, and external vibra-
tions. The dynamic transmission error model considers masses and variable stiffness of the
gear system. By considering these elements, the inertia force generates dynamic mesh reac-
tion forces. Dynamic loading in the mesh is approximated by implementing dynamic factors
that show the ratio of the dynamic load to the static load in gear design. Kayama [25] con-
clude that dynamic transmission error is equivalent to a product of static transmission error
and a dynamic transfer function. The static transmission error is a starting point for dynamic
transmission excitation. Dynamic transmission error is crucial for lightly loaded and low-
contact ratio gearing.
27

2.2.4.3 Manufacturing transmission error

As mentioned previously, any change in the tooth geometry and corresponding tooth contact
point influence on the generation of transmission error. The deflection from actual conjugate
action caused by manufacturing errors is known as manufacturing transmission error. The
assembly errors and manufacturing methods lead to positioning and profile errors that effect
on the angular position of gears during mesh cycle and generate vibration profile shown in
figure 2.13.

Figure 2.13 Example of Manufacturing transmission error. Reproduced from [14]

The long-term component is formed by the eccentricity of the gear about its rotational center
and appears as periodic vertical rise and fall in the magnitude of eccentricity error. Tooth
profile and pitch spacing errors influence on each tooth can be observed as a localized event
at individual parabolic like curves in the short-term component.

2.2.4.4 Kinematic transmission error

The derivative of manufacturing transmission error is the kinematic transmission error. Kin-
ematic transmission error studies influence of asperities like local uneven tooth surface, sur-
face roughness on deviation from true involute. The contact surface changes due to asperities
to support the loading. The geometric deviation from the actual profile causes non-desired
mesh stiffness variation and corresponding excitation. Usually, the influence of kinematic
transmission error on noise is small in comparison with the contribution of elastic and man-
ufacturing parts of transmission error.
28

2.3 Gear noise overview

Gear noise is almost always a quality issue. Force variation generates vibration in compo-
nents. Vibration transmits to the surrounding structures to excite external panels at weak
locations that generate the airborne noise. Lemanski [26] stated that excitations in gearing
are caused mainly by force variation in amplitude, direction, or position.

The difference in amplitude of contact force leads to dominant vibration in involute gears
produced with attainable accuracy. Force variation in involute gears happens due to elastic
deflection, smoothness of the drive, and the combination of small geometrical modification
that causes deviation of movement from true involute. This variation gives the variation in
response and results in vibration. Another observation shows that noise also can be caused
by oil lubricants that stuck in the roots of meshing teeth. Lubricant creates a new surface
pattern and causes time-varying force as it starts to move axially through the teeth roots.
This occurrence is common for spur gears. Oil flow is smooth in helical gears due to the
helix angle. The oil effect on noise is negligible.

Particular noise signals are a sign of poor gear design. When teeth start their contact, the tip
impact and root interference may occur. As a result, such design errors promote deviation
from the designed contact curve. Proper design of tip relief, long addendum at the pinion
and short dedendum at gear, and utilization of standard pressure angle are the solutions to
avoid particular noise signals.

This thesis work is focused on the gear whine noise caused by gear contact discussed in
Section 2.3.2. Gear sound may also be associated with gear rattle noise in lightly loaded or
unloaded gears.

2.3.1 Gear rattle

Gaps between teeth leading to lack of force and motion transition from driving wheel to
driven gear, also known as a backlash, are the main reason for gear rattle. Gear rattle is
excited by the regular force that often caused by engine fluctuation [11]. Gear stays in con-
tact under static torque. When dynamic fluctuation in torque at gear mesh exceeds static load
transmitted through gears, the loss of contact may happen. The fluctuating torque accelerates
and decelerates gears and lead to motion through clearances. Collision when teeth come back
from clearance movement into contact cause noise. Such a phenomenon is a widespread
reason for excitation in vehicle engines or machines that deal with stochastics load and have
considerable gear backlash due to design or manufacturing errors.

Mura [27] in his research work notice that separation between camshaft and crankshaft gear
teeth occurs when the peak of the phase difference due to meshing error between gear pair
match with torque fluctuations of the engine and collide back on high rotational speed. The
impact force and speed of the collision are on its maximum when torque fluctuation magni-
tude reaches its peak. In such a case degree of separation is also increases. Mura conclude
that the main gear train noise is the result of tooth impact due to gear load fluctuation caused
by torsional vibration, irregular torque from engine or rotor side, and drag forces due to seal
and bearing friction, gear sliding friction and windage.
29

It is clear to conclude that the generation of rattle noise depends on relative velocities of the
gear bodies during separation and degree of backlash between teeth. Increased static loading
at gear is the solution to reduce gear rattle noise [11]. Huang and Abram [28] confirm that
backlash is not a direct cause of rattle noise, but generated noise level seek to a minimum by
decreasing of backlash between teeth that reduce impact velocities and reaction forces be-
tween gear teeth.

2.3.2 Gear whine

The gear whine is a purely tonal sound, which tonality depends on harmonics frequency and
proportional to the speed. Gear whine is a tonal noise caused by non-conjugacy of the motion
from one gear to another. Munro and Houser [29] stated that the process of gear whine noise
is caused and influenced by transmission error, frictional and axial forces, torsional mesh
stiffness, and lubricant flow. The sound represents annoying noise on gear mesh frequency,
or it multiplies. [1]

The concept of transmission error is commonly accepted as the main parameter to describe
noise generation and discussed in detail in further chapters. Transmission error is a measure
of deviation from perfect gearing action. The deflection of components, manufacturing er-
rors, and stiffness variation cause this deviation.

Torsional mesh stiffness of pure involute gears is the ratio between the total elastic angular
deflection and the torsional load in the driving wheel. Torsional stiffness varies because of
non-constant elastic deformation due to teeth contact mechanics characterized by geometry,
number of teeth in contact, and manufacturing quality that affects the contact pattern.

From start to the end of active profile gear forces acting on a gear tooth shift in amplitude,
direction, and cause variation of bearing reaction forces. Bearings start to oscillate and radi-
ate vibration through the housing. The meshing action is a combination of sliding and rolling
of gear teeth. Any variation in sliding direction at the pitch point creates a change of friction
force direction and corresponding magnitude that may influence on vibration amplitude and
noise level.

It is possible to conclude that the main affecting factors on gear noise are manufacturing
issues and contact mechanics, that influence on deviation from actual contact pattern. In the
modern industry, manufacturing influence on noise level is reduced by appropriate quality
management. For high-quality gear units, the torsional mesh stiffness and corresponding
variation of transmission error in the gear train are the primary source of excitation and gen-
erated noise level.
30

2.3.3 Overall path to airborne noise

The noise generation starts with a combination of physical factors, tooth, and gear body
deflection under torsional loading, manufacturing errors, and drive train design flaws that
form transmission error.

The transmission error combines a combination of factors that creates a source. The funda-
mental idea is that manufacturing errors in pitch, profile, and helix may combine with gear
body distortion, tooth deflection, and external influence to create a deviation from the
smooth running of gears in mesh.

Transmission error drives internal gear dynamics via angular fluctuations to transmit vibra-
tional forces through the bearing supports. Variation of bearing forces generate air pressure
fluctuations via gearcase panels. Vibration may travel for a long distance towards the panel
to emit whine sound. The final approach of vibration from bearing housing to the radiating
surface of gearcase or external structure is challenging to validate and predict because of
damping uncertainty. The complete noise generation map is illustrated in figure 2.14.
31

Figure 2.14 Vibration excitation and transmission path


32

2.3.4 Overview of gear noise research

In recent decades, noise and vibration problem has been the focus of intense research due to
high interest from manufacturing companies and new government requirements that restrict
the acceptable level of noise power as humans and animals are more sensitive to tonal noise
[30]. J.D. Smith [1] analyze in his book numerous approaches to measurement and estima-
tion of the transmission error and present an extended description of noise and vibration
problem in terms of design and manufacturing philosophy, metrology and product develop-
ment solutions.

Experimental measurements [31] have been the focus on gear noise. The gear noise exhibits
periodic behavior. Welbourn [32] share a review on a range of researches that have been
devoted to forming a theoretical background of noise problem in the late 70s. Welbourn [32]
introduced an initial correlation between transmission error factors and gear noise. This re-
view introduced the connection between peak to peak amplitude of transmission error profile
and its influence on generated noise power. Tooth load, transmitted power, and speed of gear
train effect detected tonalities. The tooth flank modifications were specified by including tip
relief. An empirical equation has been proposed by Masuda [33] to define the correlation
between desired parameters and noise and to predict the sound pressure level of gears in
mesh. It is given in equation 2.12:

where L is the overall noise level at one meter from a gearbox, is the gear ratio, is the
transverse contact ratio, W is the transmitted power in kW, is pitch line speed in m/s and
is peak to the peak value of static transmission error. The prediction equation is valid only
for metal gears.

Many studies [28-32] have been directed towards the recognition of noise source and param-
eters to describe gear excitation. The main parameter and source of excitation is the trans-
mission error. Other studies aim to examine numerous factors to describe influence param-
eters on transmission error and formed vibration profile. Smith [1] verified that pitch errors
increase all harmonics amplitudes except mesh frequency. Welbourn [32] reported his ob-
servations on eccentricity that cause once per revolution modulation effect on tooth fre-
quency. Stephen P. Radzevich [6] defined that runout and helix errors influence on trans-
mission error profile, because of rotational axis position change that influences on loading
action on contact pattern and corresponding mesh stiffness of gear pair. Planet eccentricity
error influence on planetary transmission system has recently been described [34] to confirm
that strong harmonics variation has been detected on transmission error measurements and
caused by radial and tangential components of displacement from the ideal position.

Considerable effort has been directed towards tooth deformation from perfect involute form
under loading. Significant progress has been achieved in the understanding of static loading
contribution in transmission error and has been reported by Mark [23], who provides a the-
oretical background on influence parameters on static transmission error. An analytical
method to estimate time-varying gear mesh stiffness and the static transmission error was
33

the goal of Weber and Banaschek [3] studies. Researches and commercial software devel-
opers have successfully implemented the stiffness model for gears in contact defined by
Weber and Banaschek [3] to calculate analytically contact patterns on tooth flank and eval-
uate corresponding static transmission error.

Parametric experiments have been performed by Lin and Liou [35] to describe gear dynam-
ics' effect on contact ratio. They observed significant variation in contact patterns with pos-
sible loss of contact for low contact ratio gears. As a result, high transverse contact ratio
gears (ratio equal or high than 2.0) operate smoothly at high and low loads as a dynamic
effect decrease. The dynamic effect of damping tends to increase when the operational speed
closes to resonance frequencies.

Transmission error topic has received wide attention in the gear noise studies concentrated
on the contribution of factors that comes from design, manufacturing, and dynamic influence
parameters. Vibration excitation due to time-varying gear mesh stiffness can be assessed by
Transmission Error quantities that have emerged as practical means for noise generation.

2.4 Methods to reduce gear noise and vibration

As clarified in previous sections, the gear noise is strongly associated with transmission error
vibration profile and corresponding harmonics amplitudes. Extensive effort has been di-
rected toward understanding the robustness of the gear drive system and obtaining optimum
parameters to minimize the peak to a peak transmission error. This section presents a broad
overview of approaches to tonality free gearing, which is based on reduction and controlling
the sources of transmission error excitations.

There are no existing DIN or ISO standards that correctly specify the transmission error
parameters for quiet gear operation. A reasonable power level of noise mainly depends on
price, legal requirements, and customer expectations. The classical peak to peak transmis-
sion error magnitude is 20 for typical industry gears, where noise is not an issue. Ultra-
precision and tonality free gearing shall operate with a maximum permissible peak to peak
values of approximately 1 . [1]

2.4.1 Gear geometry approach

Noise power influence from macro geometrical design parameters such as teeth amount,
pressure angle, helix angle, pinion and gear diameters, backlash, and clearance was a topic
of several research works [1]. The main idea is to design a gear pair to maintain smooth
transition and pure conjugate action under a specific load, and the system component was
operating parameters like bearings, shaft deflection, and manufacturing allowances. For ex-
ample, in a helical gear case, the dominant effect can be caused by variation in the helix
angle due to the torsional deflection of the shaft. The system design should be directed to-
wards perfect helix matching. [1]

The definition of involute contact ratio effect on a spur and a helical gear transmission error
was a goal of several studies. Umezawa et al. [36] stated that transverse contact ratio values
higher than two are appropriate for smooth and quiet gear operation at high and low load.
Helical gear with a high total contact ratio operates quieter than spur gear.
34

The contact ratio is the ratio of contact line length along the pressure line from pinion tip to
wheel tip and the distance along the pressure tip of successive teeth. At the changeover point,
the elastic deflection changes, as well as the corresponding transmission error. Design pa-
rameters may compensate peak to peak values and significant transmission error drop. At
the changeover point of gears with transverse contact ratio equal to 2.0, the intermediate
teeth pair in contact take full design load directed along the pressure line, and both side gears
take half of the design load. It results in the lower amplitude of transmission error due to
reduced intermediate loads in comparison with classical spur gears with a contact ratio lower
than 2.0.

2.4.2 Micro-geometry of teeth approach

The general purpose of micro-geometrical modification of tooth profile is to compensate


teeth deflection by removing material from the gear tooth flanks. As a result, gear shape is
not any more perfect involute as it may include tip relief, root relief, and crowning to com-
pensate elastic deflection at changeover points. The contact properties within the designed
profile of the gear pair play a significant factor in determining the position of modifications
and outcome of the transmission error. Modifications in the tip and root of the teeth profile
are applied to eliminate high-stress concentration in corresponding areas.

The relief is usually arc-like or linear. It is sufficient to compensate teeth deflections under
load at transition points of HPSTC and LPSTC, where the most significant change of mesh
stiffness occurs. The amount of tip relief is characterized by the location of the tip or root
relief diameter, linear or rolling angle, as illustrated in figure 2.15.

Figure 2.15 Linear tip and root relief. Example of gear design with tip and root relief reproduced from [37]
35

Profile crowning and slope modifications illustrated in figure 2.16. Lead modifications are
applied to make tooth surface slightly convex and avoid high local contact at tooth ends.

Modifications of the gear tooth profiles has emerged as an efficient tool to control contact
pattern and optimize the fluctuation of mesh stiffness, that have a significant influence on
gear dynamics and corresponding excitations [37-38]. The appropriate application of micro
modifications can reduce transmission error and corresponding harmonics.

Figure 2.16 Lead crowning and lead slope modification example reproduced from [38]

2.4.3 Manufacturing quality assurance approach

Manufacturing errors are one of the significant sources of gear vibration. The gear quality is
an important parameter to assess and control. The surface roughness, tolerance assurance of
pitch errors are essential factors in reducing radiated noise. Radial run-out and cumulative
pitch errors generate long term transmission errors. Smith [1] stated that the fundamental
tooth frequency component has a significant influence on whine noise, and generated noise
is entirely irrelevant for the eccentricity effect. The eccentricity mainly gives false data on
fundamental tooth frequency and influence on the efficiency of power transmission that is
important to control. Long term transmission error appears in the sinusoidal form, as illus-
trated in figure 2.17. The individual tooth transmission error harmonics is insensible in figure
2.17 due to relatively high eccentricity displacement. The pitch errors lead to sudden drops
at the end of the tooth pair in contact and overloading, that change profile of vibration, as
presented in figure 2.18. Short term pitch error may generate amplitude and frequency mod-
ulations that can be observed as sidebands in the meshing order, as depicted in the upper part
of figure 2.18. The variation in harmonics amplitude in the frequency domain is approxi-
mately an average of pitch error magnitude around a gear. The distance between sidebands
directly depends on the period of the cumulative pitch error, as shown in the middle and
bottom parts of figure 2.18. [39] Nowadays, industry companies provide modern quality
management to preventing and reducing the influence of mistakes and defects on product
performance depending on manufacturing cost and investments.
36

Figure 2.17 Transmission error curve and corresponding spectrum of ideal gear profile with eccentricity effect
[1] [39]

Figure 2.18 Short term transmission error shape of ideal gear profile that causes amplitude and frequency
modulations (upper part), single (middle), and triple sinusoidal form of pitch error that produces different
amplitude and distance between sidebands. Reproduced from [1] and [39].
37

2.4.4 Design with load approach

The significant contribution in the topic of gear vibration was made by Harris [40], who
proposed a method to define profile modification by introducing a diagram known as Harris
Map to reduce static transmission error. The spur gear map is illustrated in figure 2.20. Harris
stated that a smooth take-over is possible when the distance between teeth in contact along
the pressure line is the same. Actual teeth elastically deflect under load and cause variation
from true involute. Harris [40] propose to represent profile modifications as a deviation from
the correct theoretical involute profile. The design of such gears gives a static transmission
error under zero loading or loading condition. The combined effect of teeth in mesh gener-
ates static transmission error to obtain pure involute and no transmission error at about a
third of the total span following the involute, as depicted in figure 2.19.

Figure 2.19 Example of Harris map for ideal spur gear. Reproduced from [41]

There are a few essential aspects to consider. Harris map has received considerable attention
in past decades due to the explanation of the handover process and loading effect on gears
in mesh. The top curve in figure 2.19 is static transmission error under zero load, and line h
assumed as a line of half-load conditions and line o represent full loading conditions. The
line around the pitch point shows the one tooth in contact regime, and lines around the
changeover point represent two teeth in contact. Ideally, the total deflection of two teeth in
contact shall correspond to the deflection of the one-teeth regime. Both contact regime ex-
pands around the changeover point and represents load influence on deformation. At curve
the effect of tip relief is compensated by elastic deflection. The gears tend to experience
the downward deflection from true involute around changeover point when applied load in-
creases. This approach allows estimating the changeover position and tip relief design pa-
rameters when final elastic deflection is known to reduce transmission error drop between
regimes and the corresponding generated noise.

In recent years, the development of the efficient method to optimize profiles has generated
considerable interest in the field of gear noise. Cai et al. [42] propose a gear design optimi-
zation method defined by the minimization of the excitation force. Since the objective is to
achieve minimum noise power, all techniques aim to create and variate profile modifications
to provide a smooth transmission error curve and lowest drop between performance regimes
due to gear contact ratio [43-44].
38

2.5 Challenges in gear design for low noise gear transmissions

A design engineer must design the gear following strength and life requirements to accom-
plish product performance requirements. The contact stress and root stress values must be
evaluated to avoid failures. It is vital to accurately assure strength issues during design for
tonality free gearing, because any design change towards gear noise reduction may dramat-
ically influence on gears life performance.

2.5.1 Contact stress

Pitting is the dominant failure mode of gear teeth. Surface fatigue failure happens due to
contact cycle repetitions with high contact stress concentration on tooth flank. It is essential
to evaluate the contact stress states precisely. Hertzian contact theory covers the assessment
method.

When two curved bodies of sphere or cylinder are in contact, the contact stress appears in a
corresponding form of point or line. This contact area deforms that results in stress, also
known as Hertzian stress, and initiate yielding.

To estimate the contact stress in gearing gear teeth are assumed as rolling cylinders. Accord-
ing to Budynas et al. [45] the contact stress can be expressed as given in equation 2.13
for all curved surfaces:

where force pressing a cylinder on a contact length and half-width is given in equation
2.14:

Equation 2.15 is a modified equation for contact pressure on cylindrical gear pairs.
39

2.5.2 Tooth root stress

Tooth root stress is the bending stress at the area that do not participate in contact. As men-
tioned previously, increasing the contact ratio reduces the overall gear noise. But the bending
strength of gears under load increases when the contact ratio value grows up [46]. It is
essential to define an optimum combination design combination between to satisfy product
requirements.

A significant effort in researching root stress problems was made by Wilfred Lewis, who
provided a formula to estimate the bending stress in a gear tooth. According to Juvinal et al.
[46], Lewis considered gear tooth as a cantilever beam with a static normal force F applied
on it. The following assumptions are valid for derived equation 2.16:

The full load is applied uniformly across the full-face width to the tip of a single
tooth at the static condition
The radial component is negligible
Friction forces are negligible
The stress concentration in the tooth fillet is not considered.

When gears are in mesh cycle, the gear contact impact happens at high velocity, so impact
factors are given in equation 2.17.

where it indicates that tooth root stress directly depends on applied tangent force corrected
by the dynamic velocity factor , overload factor , size factor and load distribution
factor , inversely affected by tooth width b, module, and tooth form factor Y. The tooth
form factor is a function from tooth shape that may vary with the change of total gear teeth
number.

Plucker et al. [47] proposed a modified calculation method for complex-shaped involute
gears. They stated that basic calculation methods, according to ISO 6336, discussed in Sec-
tion 2.7 is not accurate because the location of the maximum tooth stress deviates signifi-
cantly for different profile shifts. The contact tooth analysis performed in KISSsoft commer-
cial code firstly define load distribution, and then use standard formulas to estimate the val-
ues of tooth root stress at each segment of the tooth profile. According to KISSsoft docu-
mentation [48], the results provided by tooth contact analysis are much accurate than those
calculated fully by the standard approach, but do not assume the actual geometry of a gear
flank. It means that the tooth root requires additional external assessment following flank
gear geometry.
40

2.6 Gear simulation models

This section presents key aspects of proposed models to predict the dynamic behavior of the
gear systems and discuss correlation with transmission error. The literature contains different
approaches that all agreed to consider torsional resonance, cyclic fluctuations, transmission
error, system component vibration response as a source of vibration in the gear system.

The principal concept of all proposed models is to represent a vibrating system made of two
gears and linked by mesh stiffness spring. Tuplin [49] presented the first mass-spring model
of gears in contact. Then Gregory et al. [50] performed intensive work on the improvement
of the proposed model. The spring stiffness represents mesh stiffness that formulates a dy-
namic model of a gear pair.

Lately, Yang et al. [51] define two different models with tooth friction, Hertzian damping,
and energy dissipation to predict vibrations of gears in mesh. Non-linear aspects of dynamic
models have been the focus of intense research for several recent decades. Singh et al. [52]
studied the effect of varying mesh stiffness on dynamic behavior. The mathematical model
was extended by Burak et al. [53], who considers the effect of shafts and bearings dynamics
into account.

2.6.1 Mathematical modeling of the gear system

The mathematical models have a torsional one-degree of freedom limited by the requirement
of predefined mesh stiffness. The simple mathematical model of gears dynamics in mesh
was proposed by Gregory et al. [50] in 1962. The gear mesh model is two rigid disks that
include the equivalent masses of each gear and corresponding total moment of inertia, and
spring with predefined varying mesh stiffness. A two-degree-of-freedom model of spur gear
pair is illustrated in figure 2.20. The spring represents the torsional flexibility of gears in
contact. Spring connects the base circles of each respective gear, and it acts along the line of
action of the two gears. The spring mesh stiffness at this model is a time-variant continuous
function, which depends on the number of gear teeth in contact. Other elements of the gear-
box are assumed to be rigid.

Figure 2.20 The configuration of gear contact model proposed by Gregory et al. [50]

Transmission error for the gear drive is a linear displacement along the line of action.
41

The advanced model is illustrated in figure 2.20 and was developed by Singh et al. [54], who
analyze gear contact with the assumption of viscous damping. The time-variant mesh stiff-
ness includes an assumption of the backlash effect.

Figure 2.21 Advanced gear contact model proposed by Singh et al. [54]

Mathematical modeling of gear drive response including shaft, housing, and bearing dynam-
ics with non-linearities coming from clearance and preload has been the focus of several
papers[55-57]. The non-linear multi-degree-of-freedom model includes shafts with the lat-
eral stiffness, damping parameters, moment of inertia and bearings, that considered as
springs. The beams can be assumed as continuous flexible beams. Torsional vibration can
be reconstructed by adding springs at bearing points to the model torsional stiffness of shafts.
Corresponding validation is an essential part of the assessment of described analytical mod-
els. Kahraman et al. [58] experimentally studied the effect of shaft flexibility on the spur and
helical gear dynamics by assessing transmission error and axial motion. They stated that
changes in shaft flexibility generate additional natural modes and that helical gears exhibit
additional axial motions and rocking effect. Kahraman et al. [58] also extended their research
by including a sinusoidal or a periodic mesh stiffness to study interactions between the mesh
stiffness variation and bearing clearance. They concluded that bound between variation in
mesh stiffness and bearing nonlinearities is negligible. However, this report justifies that
existing mathematical torsional models cannot be applied to represent actual applications.

The Finite Element Method has emerged as an attractive alternative for dynamic modeling
of complex continuous gearing systems, including non-linearities discussed above.
42

2.6.2 Non-linear finite element approximation of the gear system model

A high computational cost finite element-based solution is the main alternative to analytical
models due to its adaptivity to complex gear drive systems, manufacturing errors, gear body
geometry, irregular shapes of tooth face, and gear rim thickness, that usually ignored by
analytical mathematical models. However, the significant advance of the finite element
method over the most analytical models is the elimination of the requirement to determine
initial mesh stiffness.

The Finite Element method has emerged as an effective solution to simulate varying contact
conditions and assess the dynamic behavior of complex gear systems. Gurkan [59] modeled
a hybrid finite element model with the lumped parameter, which represents a periodic linear
displacement of predefined static transmission error. The developed model is capable of cal-
culating the dynamic transmission error (DTE) and bearing forces to assess the modal and
acoustic response of the transmission system. A static transmission error function for the
spur gears is straightforward to define and implement, which makes this model correlate
well with experimental studies. However, transmission error function for helical gear is dif-
ficult to predict and implement in such a model.

Advanced researches [60-62] focused on the contact mechanics of the geared system with
the full FEM approach. Hence there is an advantage in improving the accuracy of response
prediction due to the capability of modeling complex gear systems without input parameters
of initial static transmission error or mesh stiffness function. In this case, the angular velocity
of the driving gear and torque on the driving wheel usually is assumed to be constant, so the
angular velocity difference between input and output points of the system is used to estimate
static transmission error as given in equation 2.10.

Detailed finite element modeling of the gear system and components have been the topic of
intensive research [63-65] that successfully validates reference models' performance over
mathematical contact model and present advance models accounting shafts, and bearings
support as well as the tooth compliance. Allison and Hearn [66] performed a comparison
between the experimental and finite element models of gears in contact and received a good
correlation of bending stress values between both cases.

Contact condition is a unique discontinuous constraint on touching bodies. The FEM code
shall automatically define a contact area to apply constraints through which the forces allow
to be transmitted from one body to another and to detect when separation occurs to remove
constraint conditions. It is a challenging task due to presented non-linearities from the ma-
terial, geometry, and contact motion implemented around the Newton-Raphson method. The
state of all contact iterations is checked at the start of each increment by the algorithm. As
result nodes, the pair's contacts are either open or closed. A constraint condition is assigned
to closed node pairs. During the iteration, the process solver updates the configuration of the
model according to the contact condition. After each iteration, the clearance and contact
pressure between nodes are checked. If the clearance is detected to be negative or zero, the
nodes' status becomes closed, and new constraints are applied. If the contact pressure is de-
tected to be negative, the nodes' status becomes open, and constraints expire. The iterations
are converged when no changes in contact status occur. Finally, the cycle continues until
convergence occurs. Parker and Viandyaker [67] used the described above principle to create
a two-dimensional finite element model of spur gears and to obtain results of contact losses,
43

mesh stiffness, and static transmission error. One of the outcomes of this thesis is to imple-
ment this approach on three-dimensional helical gear teeth with flank modifications.

Another category of finite element-based researches concentrates on the influence of gear-


box elements on dynamic response. Velex et al. [68-69] establish a model where the contact
problem was solved along with a discretized contact pattern on a teeth surface for each time
step and was coupled to dynamic mathematical models represented by classic shafts and
lumped parameter elements. The proposed solution allows us to solve mathematical models
without predefined mesh stiffness function and to account for gear teeth shape with the effect
of dynamic response from gearbox elements, including shaft and bearings. Perret-Liaudet et
al. [70] coupled a finite element model of gear components with the mathematical lumped
model to study the influence of gearbox component stiffness on dynamic behavior. The au-
thors stated that the elastic properties of components effect obtained response results and
thus justified requirement for accurate stiffness representation of the gearbox components.

Last group aimed to analyze full gearbox vibration and related sound field. Zhou et al. [71]
assessed the vibration modes shapes of industry gearbox using hybrid mathematical models
of gears in contact. They analytically define dynamic bearing forces and use them as an
excitation mechanism for vibration analysis of the finite element model of housing. Finally,
they implemented hemispherical sound fields to define weak points of gearbox casing to
describe possible transition paths to the radiating surface that allows proposing approaches
for minimizing of the sound pressure level. This type of analysis is impossible to perform
via analytical tools. Guo et al. [72] performed gearbox vibration analysis and successful
experimental validation on a complete finite element model of the gearbox, including contact
mechanics of simple spur gears without modifications, rolling, and supporting elements.
They did not consider the contact of rolling elements in bearings, neglected the effect of
inertia and discretized the gear contact pattern into the distribution of linear springs to reduce
computation cost. The authors proposed to predict bearing force reactions based on contact
analysis and use as input data for noise pressure prediction from the housing model. They
conclude that experimental noise measurements correspond well to results obtained via the
finite element model approach. Finally, they concluded that the finite element approach al-
lows to estimate gearbox vibration accurately and to perform optimization studies to reduce
noise radiation as it allows to accurately model influence of boundary element on the hous-
ing.
44

2.7 Analytical gear teeth contacts analysis with KISSsoft

As stated, the variation of the tooth meshing stiffness previously is a primary source of gear
excitations. The determination of meshing stiffness in a correct and precise way is an essen-
tial aspect of the analytical contact analysis model implemented in KISSsoft AG and used
by Moventas Gears OY. KISSsoft is a software for creating, optimizing designs, and calcu-
lating corresponding parameters for machine components of powertrain transmission sys-
tems. The gear module of the KISSsoft software is responsible for formulation contact anal-
ysis between teeth under pressure, including the influence of macro and micro design fea-
tures. The primary purpose contact analysis is to evaluate the gear mesh under load and to
establish an analytical tooth stiffness model based on the load distribution, the tooth defor-
mation, shaft, and support performance and manufacturing errors. The method is based on
the tooth deflection during meshing action under loading. The deflection consists of three
components. At first, the tooth surface deforms. Second, the gear body bends. Finally, tooth
contour flattening caused by Hertzian pressure occurs in the contact area. The analytical
method assumes the following assumptions:

the bodies are isotropic and homogeneous


the dimensions of the contact area are microscopic compared with the radii of curva-
ture
the contact area is completely smooth
no shear stresses presented in the interacting surfaces
the contacting surfaces behave linearly elastically
Hertzian theory assume that contacting surfaces are frictionless

The next sections outline the details and essential theoretical aspects used to perform contact
analysis in gear design. The contact analysis consists of two linked analyses. Section 2.7.1
describes an analytical method to analyze load distribution across the face width. And sec-
tion 2.7.2 provides a method to analyze the actual behavior of tooth pair under loading in
terms of mesh stiffness and transmission error.
45

2.7.1 The load distribution along the face width

At first, the load distribution on the path of contact required is vital to assess. The load dis-
tribution factor characterize the face load. The calculation of load distribution along the
face width is made according to ISO6336-1 Chapter 7 Annex E. Method specified in ISO
6336-1 Annex E assumes non-linear load distribution, consider the influence of shaft deflec-
tion, torsional deformation, support components, and manufacturing allowances effect on an
unequal tooth load distribution on the contact stress, tooth flank ( and tooth root stress
( . Axis alignment, manufacturing imperfections, bearing performance, elastic defor-
mation of drive chain components lead to uneven load distribution along tooth face width.
The face load factor provides data to review the contact pattern to determine the optimal
tooth trace avoiding edge contact and compensating shaft and bearing influence.

A gear contact model was developed and implemented in KISSsoft code to perform calcu-
lations according to ISO 6336-1 Annex E. The model is illustrated in figure 2.22. The best
way to simplify the iteration process to solve the system of equations of a complex mathe-
matical gear model is to sub-divide the gear teeth into equal n number of slices. Each slice
examines actual tooth trace modification and deformation to estimate the actual mesh gap.

Figure 2.22 Tooth flanks of length b and misalignment are brought together at common contact width
under three cases: load free (at left), small load (center), and large load (right). Reproduced from [4]

Spur gear slices link together to replace the helical gear's transverse section of each tooth.
Links represent the spring system. An example of helical gear representation as a slice spur
gear model is illustrated in figure 3.7. Slice number depends on the accuracy setting defined
by the user. According to the KISSsoft user manual [5], increasing the number of slices
improves the quality of results but significantly increases computational time, requirements
to internal operative memory, and available capacity on a hard disk.

The face load factor is proportional to contact stiffness and inverse proportional to average
load on a contact line, as given in equation 2.18. Mesh stiffness is determined ac-
cording to Weber and Banaschek [3] and discussed in the next section.
46

The calculations concentrate on component effect on elastic deflections of the pinion and
wheel shafts following ISO 6336-1 without considering elastic deformation of housing, bear-
ings, and effect of bearing clearance. It considers the effect of gear drive elements arrange-
ment concerning the transmission path and effect of manufacturing errors to axis misalign-
ment according to ISO/TR 10064-4 that includes out of plane errors and inclination errors.
Manufacturing errors follow the normal distribution, and 99.7% of gears are produced within
specification, so the 3-sigma rule can be used to obtain manufacturing mesh misalignment.
Examples of the line load distribution are shown in figure 2.23 and figure 2.24.

Figure 2.23 Example of the line load distribution across face width for spur gear

Figure 2.24 Example of the line load distribution across face width for helical gear pair
47

2.7.2 Loaded tooth contact analysis

The load distribution is an essential component in the tooth deformation analysis along the
path of contact. Weber and Banaschek [3] established the method to predict contact behavior.

The fundament of mesh stiffness theory, according to Weber and Banaschek [3], states that
total gear deformation in mesh as given in equation 2.19:

where three main components are bending , tilting and Hertzian flattening

The contact analysis principle algorithm is illustrated in figure 2.25.

Figure 2.25 Principle algorithm of contact analysis

According to Weber and Banaschek [3], elastic beam and plate theory are the basement for
gear deformation calculation. It is essential to use the state of plane strain model since the
models of spur slices at contact points are infinite in perpendicular to the driving direction
and undergo loads in the base tangent plane.

The contact points are determined in a teeth model for each mesh position at the time instance
. The contact pattern along the tooth is discretized to calculation points in two-dimensional
slice models. Utilization of analytical plate deformation theory allows determining defor-
mation components under bending, tilting, and Hertzian flattening to obtain global mesh
stiffness and total deformation. An orthogonal cartesian coordinate system is the origin of
the model presented in figure 2.26.
48

Figure 2.26 Analytical model of gear tooth deformation under bending (top left), tilting (top right), and Hertz-
ian flattening (bottom). Reproduced from [48]

Equation 2.20 describes the bending deformation of the mechanical substitutive two-dimen-
sional teeth model in mesh illustrated in figure 2.26. The first deformation term inside the
brackets is responsible for effect from strain energy of bending moment, and the second term
provides the effect of strain energy of shear and normal force.

(2.21)

Equation 2.21 describes the tilting deformation of the mechanical substitutive two-dimen-
sional teeth model in mesh illustrated in figure 3.8. The first deformation term inside the
brackets is responsible for effect from strain energy of bending moment, the second term
49

provides the effect of strain energy of shearing force, and the last term generates influence
from strain energy of normal force.

According to Hertzian theory [73], cylinders, balls, or ellipsoids pressed together represent
teeth form. Helical gear contact analysis uses the truncated cone form. When two curved
surfaces interact, the flattening occurs. The area of contact has a line-shaped form. The
Hertzian theory is an efficient way to obtain contact stresses and to analyze the size and form
of the compressed contact zone. When two cylinders are in contact, the surfaces normal in
the contact point are orientated in the direction of compression and parallel to each other
[73]. It allows estimating the normal distribution of pressure in a plane of action as presented
to obtain the flattening deformation of the two bodies. The Hertzian flattening is given in
equation 2.22.

The deformation terms inside the brackets of the equation are responsible for Hertzian flat-
tening of two substitutive teeth.

Finally, the total deflection is a sum of corresponding components of equation 3.18. Bending
and tilting deflections are linear, but Hertzian deflection is non-linear. The corresponding
stiffness from force and interference of each meshing tooth pair are given in equations 2.23-
2.25.

The final contact mechanics schematic diagram of gear mesh is illustrated in figure 2.27 for
one slice and figure 3.28 for complete teeth meshing with several slice models. The tangent
contact stiffness in the operating condition is given in equation 3.26 and correlates well
with given in equation 2.27 determined to obtain the face load factor.
50

Figure 2.27 Discretized contact model schematic diagram of one slice. Reproduced from [4]

Figure 2.28 A discretized teeth model for meshing with several slices to represent full gear contact of helical
gear pair. Reproduced from [74]
51

The single teeth contact stiffness is the serial combination of each spring stiffness represent-
ing bending, tilting, and Hertzian, as illustrated in the upper part of figure 2.27. The slice
mesh stiffness model represents the system of parallel springs of teeth pair in mesh with
corresponding contact stiffness depicted at the bottom part of figure 2.27. The teeth are di-
vided into slices across the width, as illustrated in figure 2.28. Numbers of slices must
return resulting torque equal to nominal torque. Slices are coupled parallelly together by
torsional coupling stiffness defined in equation 2.28 to provide accurate data and transfer
the influence of neighbor slice elements.

The 0.04 is an empirical coefficient derived by KISSsoft AG comparative studies [48] in the
finite element contact analysis. This coefficient can be modified case manually by case. Root
stiffness is the serial combination of bending and tilting mesh stiffnesses. It is important
to note that there is no coupling of the Hertzian deformation between the slices.

Finally, the resulting torque is equal to the contribution of all contact points and slices
and is given in equation 2.29:

where the contact radius is used instead of the base radius and determined from the con-
tact point
defined by the engineer. The solution converged when the resulting torque is compared with
the nominal torque, as given in equation 2.30:

When the iteration is converged, the wheel is twisted against pinion to a new calculation
position. Tooth deflection and mesh stiffness are the essential variables to assess noise gen-
eration. The linear displacement on a contact path corresponds to the transmission error.
Other factors and performance characteristics are derived from the results of Weber and
Banaschek [3] calculation and covered in the following sources [75-77] and [48].
52

2.8 Evaluation of affecting factors on analytical solution results

This section introduces parameters that may affect the efficiency of the method implemented
in KISSsoft while evaluating the overall operational behavior. Contact analysis discussed in
Section 2.7 provides a wide variety of gear performance characteristics that includes contact
ratio under load, transmission error characteristics, tooth pair stiffness, load distribution,
Hertzian pressure, torque curve, contact lines, and other results that are not necessary for this
study. When helical gears with the sophisticated modified design are examined, specific af-
fecting parameters should be considered during results interpretation. Influence factors ob-
served during thesis performance are caused by simplifications and assumptions used by
Weber and Banaschek [3] and implemented in KISSsoft to make the analytical mathematical
model more amenable to calculation.

At first, Winter [75-76] has observed the difference between measured and estimated Hertz-
ian deformation for spur gear. It was also stated that bending and tilting components are in
good agreement. Winter proposes to apply a correction factor for the Hertzian
deformation to normalized results.

Secondly, as stated in previous sections, the contact stiffness of helical gears always varies
less then spur gear due to contact ratio. Calculated bending and tilting stiffness at the end of
the face width is less than nominal due to decreased tooth thickness at the edges. It causes
buttressing effects presented as the high peak of stresses and normal force at the start and
end of the mesh. An example of thickness cuts in the normal direction is illustrated in figure
2.29.

Figure 2.29 Illustration of two cuts for gear with helical teeth, where is nominal thickness and is
reduced thickness. Reproduced from [4]

An effect of reduction in bending and tilting stiffness is equal to the product of stiffness
and border weakening factor as given in equation 2.31:

An exponent value of 0.5 was derived in comparative analyses [78]. This factor has a signif-
icant influence on the representation of the buttressing effect that occurs in helical gear teeth.
The border weakening factor can be manually modified case by case. According to the trans-
mission error theory, significant normal force variation increases peak to peak amplitude.
53

The stiffness results deliver the stiffness curves of tooth pair in mesh along the path of con-
tact. The system stiffness is the torsional stiffness of the gear pair and depicted in figure
2.30.

Figure 2.30 Example of spur gear system stiffness defined by tangent stiffness and from estimated
function of deflection and the nominal transverse load

Torsional stiffness of gear pair consists of tangent stiffness and secant stiffness .
Those stiffnesses correspond well to theoretically estimated meshing stiffness and
which are the mean stiffness for the unit face width in ISO 6336-1. The correction factor for
Hertzian stiffness and corresponding deflection has a considerable effect.

The mesh stiffness calculation is made with different Hertzian factors on the standard spur
and helical gear according to DIN3990 to depict changes and illustrate influence in figure
2.31. The mesh stiffness profile shifts in peak to peak amplitudes, shape, and magnitude due
to difference in Hertzian deformation. Measured Hertzian flattening is higher than calculated
according to Weber/Banaschek. Winter [75-76] performed validation on standard helical and
spur gears. The determined coefficient is implemented in KISSsoft calculations. It is essen-
tial to define and validate the correct Hertzian factor experimentally for helical gear with
teeth modifications.

Nevertheless, better convergence occurs for high contact ratio gear designs, where the influ-
ence of the bending is dominant, and the influence of the Hertzian deformation is minor. The
influence of the Hertzian factor variation on helical gear is presented in the lower part of
figure 2.31. According to the Hertzian theory [73], the distribution and magnitude of Hertz-
ian pressure depend on the normal force and the local curvature radius of both flanks at the
contact points. It is essential to control the peaks of normal forces at the SAP point and EAP
point, as shown in figure 2.32 and known as buttressing effect. The contact line length varies
during one tooth pair mesh period, and load distribution is nonuniform due to the variation
of teeth in contact at LPSTC and HPSTC points. Reduced bending stiffness at the helical
tooth edges creates the buttressing effect on SAP and EAP contact points.
54

Figure 2.31 Influence of correction factor on the spur and helical gear mesh stiffness

Figure 2.32 Buttressing effect of helical gears

The gear teeth modifications also affect obtained results. When tooth profile design includes
linear tip relief, the corresponding normal angle change hit cause stepped curved disturb-
ances on the normal force distribution curve. The stepped curved is illustrated in figure 3.23
and caused by instant variation of the normal angle between pure involute profile and tip
relief modification. Figure 2.33 depicts the influence on Hertzian stress of instant transition
from the correct involute profile to tip relief with zero curvature radius, which is not usually
realistic due to manufacturing technique and tools. The Hertzian stress generates small wav-
iness fluctuations at a normal angle change position on the curve. Manufactures avoid the
immediate change of normal force by adding arc-type or progressive type of tip relief with
transition radius.
55

Figure 2.33 Effect due to linear tip relief

As mentioned in previous sections, the excitation is formed by variation in force, correspond-


ing amplitude, direction, and position of action. It affects contact stiffness and corresponding
vibration profile. Transmission error is the tooth deformation defined by displacement of
contact point from the correct path of contact.

Design parameters, modifications, and discussed above parameters formed transmission er-
ror vibration profile and cause a significant effect on the estimated spectrum diagram. Mean-
while, the peak to peak values is identical, as illustrated in figure 2.34.

Figure 2.34 Example of Spectrum analysis of transmission error profile with stepwise function (upside) and
smooth transition function (bottom). Reproduced from [4].
56

When the transmission error profile is a stepwise function, the obtained spectrum data char-
acterize a spectrum with large amplitude values in the higher-order harmonics. But when
transmission error profile is a smooth function, the higher-order terms disappear. According
to Klingelnberg [39], the higher-order harmonics decrease exponentially, when transmission
error is smooth and without waviness, meanwhile waviness caused by kinematic transmis-
sion error and other design or modeling imperfections, shows jumps in higher-order harmon-
ics. An example of this observation is illustrated in figure 2.35.

Figure 2.35 Example of transmission error curve without and with waviness. Reproduced from [39]
57

3
This chapter presents the FEM approach used to estimate characteristics of examined gear
system excitations in terms of corresponding static transmission error of gears in mesh. It
provides an approximate solution to represent a complete gearbox performance model, in-
cluding gear body geometry factors, gear rim thickness, and cross-influence from related
components. Section 3.2 discusses the comparison analysis of both methods to find out how
the obtained results correlate with each other and present assessment criteria to define the
range of applicability of selected methods.

3.1 Gear teeth contact analysis using finite element method

This section explains the method developed in this thesis work based on FEM. The exact
simulation of gear teeth contact is a challenging task and requires high computational efforts,
as stated previously. The challenge is to achieve the state of gear contact in a precise way.
The FEM is used to approximate deformation, static transmission error, mesh stiffness, con-
tact pattern, and stress distribution. The gear contact model procedure relies on the finite
element approximation of teeth and deformation contact model considering the effects of
gear body geometry, shaft deflection, bearing, and housing supports using ANSYS Work-
bench tools. Further sections introduce modeling procedure details. Section 3.3.5 discusses
the validation and verification of finite models.

3.1.1 Geometry

The solid geometry has been constructed using CATIA V5R27. The three-dimensional finite
element model includes components of interest to enable the contact analysis during one
mesh cycle and allows dealing with complex load distribution on the geometry of helical
gear teeth. The model consists of shafts, pinion and wheel, bearings and housing supports
constructed according to the Moventas's Exceed Series high-speed helical gear stage tech-
nical documentation. The content of technical documentation is the proprietary and confi-
dential information of Moventas. The primary data and corresponding dimensions of con-
structed for this thesis work pinion and wheel gears are provided in Appendix 3.

The reference gearbox examined in this thesis work was designed for deployment in wind
power generation and shown in figure 3.1. It has the function to transmit power delivered
from the rotor side to the electricity generator side, to increase rotational speed and decrease
the torque. It is a megawatt power class gearbox to perform in up to 4 MW wind power
turbines. The gear ratio per single reduction is limited, so there are three stages: two plane-
tary and one parallel presented to obtain gear ratio 1:160. The examined helical stage, illus-
trated in figure 3.1, is the last stage of power transmission. The construction consists of two
shafts with attached helical gear and supported by rolling bearings. Assembly is covered by
housing. The input shaft of the helical stage is an intermediate shaft depicted as a hollow
shaft in figure 3.2. An intermediate shaft connects the second planetary stage of gearbox and
driving wheel of the helical stage and transfer power to the high-speed shaft. The high-speed
shaft is the output shaft and connected to the generator coupling, that is not shown and con-
sidered at this thesis work. All bearings are single row type. The rotor side bearing on the
intermediate shaft is a cylindrical roller bearing. Other bearings are taper roller bearings
used in duplex pair to carry dual-acting axial and radial loads.
58

Figure 3.1 Exceed Series Moventas Gear OY Wind Gearbox

Figure 3.2 Simplified illustration of the helical stage examined in the thesis work. The left side corresponds to
the general representation of helical stage housing from the generator side and right part depicted inside the
assembly.
59

Figure 3.3 Simplified representation of examined helical transmission train that consists of the driven gear,
driving wheel, bearings, and shaft models. The left side of the figure is the rotor side, and the right side is the
generator side

The gear train increases rotational speed and reduces the torque with a speed ratio of
. Speed ratio, also known as gear ratio, indicates that driving gear rotates 4.43 times
slower than driven gear. Both gears have the same helix angle magnitude of , but
oppositely oriented. Driven helix angle cut with a right-hand orientation. The high-speed
shaft contains integrated gear teeth and connected to intermediate shaft via gear
wheel with gear teeth. The corresponding contact ratio is . It indicates that
three teeth have contact 56% of mesh period, and four teeth are in contact other 44% of mesh
period. The gear teeth include modifications of arc-like tip relief, helix angle modification,
crowning, and arc-like end relief of size provided by Moventas. Other components associ-
ated with helical stage assembly are not included in the analysis due to neglected influence
on the thesis research objective.

3.1.2 Material characteristics

The material of the housing frame is spherical cast iron GJS-400-18-LT (SFS-EN 1563) the
shaft material is 42CrMo4, pinion, and ring wheel material is 18CrNiMo7-6 and bearings
material is structural steel. Table 3.1 introduce material constants according to SFS-EN
1563 and SFS-EN 10083-3.

Table 3.1 Material properties - is the yield stress for tensile and is the yield stress for compression,
UTS - ultimate strength

[ [MPa]
GJS-400-18-LT 7100 169 0.275 200 260 285
42CrMo4 7850 210 0.3 454 -454 730
18CrNiMo7-6 7850 206 0.3 850 850 1060
Structural steel 7850 210 0.3 250 -250 460
60

3.1.3 Finite element mesh models

The FEM-model has been analyzed and post-processed in ANSYS 2019 R3 software. The
finite element mesh of the full model includes 1107598 nodes and 4097836 of elements. The
mesh density and element quality are set higher on contact surfaces. The mesh consists of
linear tetrahedrons, hexahedrons, wedge and pyramid elements as well as quadratic contact
elements. A detailed description of the developed finite element models is presented below.

3.1.3.1 Gear teeth finite element model

The gear geometry construction process starts with a unique formulation algorithm of
KISSsoft software to assure precise involute profile (deviation from the actual profile is up
to 1 . The capability for setting up the desired accuracy is implemented in the algorithm.
KISSsoft uses the same formulation principle as for contact calculation and generates helical
gears using layers of spur gears connected by linear and spline interpolation curves along
with the involute profile, as illustrated in figure 3.4. Due to this reason, contact may occur
nonuniformly in finite element software due to deviation of surfaces from true involute. The
same number of sections was used for the formulation of the KISSsoft calculation model
and FEM model to assure accurate representation of designed geometry and provided results.

The 75 sections allow reaching the desired accuracy for the robust model with modeling
Finally, the Stand-
ard for the Exchange of Product Data (STEP) format [74] accepts to transfer data into FEM
software.

Figure 3.4 Solid parametric teeth model formed in KissSoft, where splines connect vertices on spur sections,
and point resolution increases at modification points.

It is adequate to obtain data for one mesh cycle to perform Fast Fourier Transform analysis
of static transmission error. There is a large number of teeth that do not participate in one
cycle contact and have a negligible influence on contact forces and system mesh stiffness.
There are only 7 teeth per gear, as illustrated in figure 3.5, to reduce computational cost and
complete one mesh cycle. Teeth are separate parts to implement the desired finite element
meshing technique.
61

Figure 3.5 High-Speed Shaft simplified initial model at the left side and driving wheel simplified model at the
right side.

Few affecting factors affect mesh selection and construction. The contact is localized on a
very narrow strip on teeth with non-conformal surfaces. The obtained contact forces act on
a tiny area, that may cause computational errors in stress and deformation data at a distance
of one to two elements [3] away from narrow concentrated zones due to principles of finite
element analysis, so the dense mesh is required to obtain accurate results. The contact stress
distribution example for the fine and coarse grid is illustrated in figure 3.6. The accuracy of
stress and deformation states are essential for correct gear performance simulation of the
physical gear pair model. These inaccuracies cause the prediction of contact force distribu-
tion along the teeth surface to be erratic [3]. Corresponding uncertainties affect transmission
error. As mentioned in Section 2.2, the transmission error is formed by variation in force,
corresponding amplitude, direction, and position of action.

Figure 3.6 The contact stress distribution. The left-side delivers the contact line stress distribution on refine-
ment mesh; meanwhile, the right-side shows imperfections caused by the coarse mesh. There is bending stress
influence presented for both cases and depicted as the blue area below contact lines.

Another affecting factor results from contact force sensitivity to geometry modifications of
the teeth. These surface corrections are small-scaled and have a noticeable impact on gear
contact performance as the normal force acting on teeth changes direction and influence on
62

contact force distribution. The finite meshing for gear profile is geometrically accurate at
places of modifications and sensitive to the effect of surface modification to satisfy the gear
design functionality.

The finite element teeth pair consist of linear three-dimensional continuum four nodes hex-
ahedron and six nodes prism elements. Sizing functions control the growth and distribution
of mesh elements in the places of high curvature at tooth edges. This function determines
the tooth face sizes based on edge curvature normal angle and makes mesh to be geometri-
cally precise at the modification transition coordinates. Span angle center was defined as
fine to refine curvatures at tooth surfaces., the inflation option generates a thin layer along
the tooth surface to resolve high-stress concentration and to minimize the appearance of
distorted elements in the mesh. This function adjusts the hexahedron and prism cells in the
layer according to tooth surface boundaries. Mapped face meshing option creates structured
mesh on tooth surfaces.

Mesh dependency studies were performed to define optimum size, exclude finite mesh in-
fluence on stress distribution. The number of mesh divisions across the face width region
was changed. The tooth pair contact analysis was performed to obtain deflection and stress
states. Stress distribution behavior at contact and fillets from coarse mesh to fine mesh cor-
respond to observations made in figure 3.7. The deflection is the transmission error. Results
presented in figure 3.8 denote that dense mesh is required. The relatively small deflection
change occurred when mesh density was increased. Finally, the total number of elements
selected is elements and the corresponding number of nodes is 7 . The finite
element model of gear teeth for the high-speed shaft and driving wheel is illustrated in figure
3.24.
63

Figure 3.7 Mesh dependency study of linear deflection.

Figure 3.8 Finite element model used for gear teeth. The upper part of the figure illustrates the distribution of
mesh across the high-speed shaft teeth surfaces. The bottom part of the figure depicts the mesh distribution
across the driving wheel teeth.
64

3.1.3.2 Pinion and wheel ring finite element model

This section introduces the finite element model of assembly that consists of teeth discussed
in the previous section, the wheel ring, and pinion. The shafts introduce system compliance
that leads to misalignment and affects the excitations. The system finite element model is
illustrated in figure 3.9. The model represents the actual geometry of the industry gear drive.
The entire system was constructed with the linear tetrahedron, wedge, and pyramid elements.
The model contains a hollow shaft input, also known as the driving wheel. The merged to-
pology between hollow shaft and gear teeth bodies allows achieving a conformal mesh where
body interaction occurs. The first teeth contact position corresponds to the HPSTC contact
point. Sizing functions assure a smooth transition and control the distribution of mesh ele-
ments in the places of chamfers and fillets. The total number of elements constructed is
elements and the corresponding number of nodes is 8 .

Figure 3.9 The overall configuration of the model is illustrated at the left side of the figure, including the input
shaft, driving wheel, and driven shaft. The gear teeth finite element models merged with the high-speed shaft
and the driving wheel are illustrated at the right part of the figure.
65

3.1.3.3 Finite element modeling of rolling bearing supports

In the actual gear drive system, bearings support shafts from the housing. Kinematic con-
nection of bearings is computationally challenging to approximate efficient, because of com-
plex bearing rolling elements dynamics, that require additional efforts to model bearing ele-
ment contacts and clearance.

An equivalent parametric model of bearings is introduced to reduce computational cost and


make it possible to perform this study with selected tools. Bearing consists of an assembly
of two coupling rings corresponded to actual geometry. Bearing rings are constructed with
tetrahedron elements. Inner coupling is perfectly bonded to shaft surfaces and connected to
outer coupling by bushing joint spring. Spring represents bearing stiffness matrix at the op-
erating point. A bushing spring has three translations and three rotations degrees of freedom
that characterized through their corresponding translational and rotational stiffness. The
stiffness matrix is based on mechanical loads, shaft deflection, and influence from the elas-
ticity of rolling contacts in gear drive system from an integrated database of BEARINX
software that used for selection and analysis of rolling bearings arrangements according to
ISO/TS 16281 and DIN 26281.

The bearing loads in KISSsoft and BEARINX is estimated from the applied forces. Nominal
torque ( ) is applied to the pinion, and the rotational velocity is
defined for the driving wheel. The reaction forces calculation process is defined in ISO/TS
16281 and implemented in KISSsoft and BEARINX software. Boundary conditions used for
FEM evaluations are illustrated in figure 3.16. Joint probes detect reaction forces for FEM
evaluation. The bearing convergence is evaluated by comparing desired static reaction forces
for selected supports with KISSsoft and FEM solutions. The obtained results are presented
in Table 3.2 at the HPSTC contact point when the arc of contact length is on its maximum.
The computational model used to approximate the gear drive system with bearing supports
and bearing the legend map is illustrated in figure 3.10. There are taper rolling bearings T1
and T2 installed on the high-speed shaft. And there is a tapered rolling bearing T3-T4 pair
and cylindrical rolling bearing R1 installed on the intermediate input shaft.

The difference in geometry representation explains the divergence in obtained results be-
tween KISSsoft and BEARINX. The non-linear bearing stiffness provided by BEARINX
software is nonsymmetric, which may cause the observed difference between BEARINX
and FEM approximation, as bushing joint spring stiffness is symmetric. The axial deflection
of the high-speed shaft at the operational point obtained by the finite element solution cor-
responds well to the analytical solution, but overall deflection behavior illustrated in figure
3.11 shows the difference in a parabolic shape. It is possible to conclude that the bearing
modeling way is not accurately replicating the load paths and displacements found around
the bearing. This bearing modeling approach may reflect the overall results of contact pass,
stress states, and transmission error.
66

Table 3.2 Reaction forces obtained according to ISO/TS 16281 (BEARINX), KISSsoft and FEM; T1-T4 taper
rolling bearings, R1 - cylindrical rolling bearing.

BEARINX
T1 111.3 -126.8 34.6 679.571 3547.679
T2 -38.6 -117.2 64.7 -2502.362 -4511.034
T3 0 0 0 0 0
T4 91.4 116.9 48.2 2003 -5684.091
R1 0 129 -147.6 -81.56 -43.545
KISSsoft
T1 120.2 -125.6 34.6 1146.172 4393.258
T2 45.8 -118.4 64.7 -2406.5 -4375.5
T3 0 0 0 0 0
T4 91.3 122.5 49.3 2178.2 -5928.48
R1 0 131.8 -148.7 -47.173 -26.398
ANSYS
T1 106 -135.3 36.7 791.4 3997.3
T2 -37.3 -112.5 59.4 -2217.8 -4244.4
T3 0 0 0 0 0
T4 93.9 108.3 47.0 1662.6 -5528.5
R1 0 136.5 -147.2 -139.8 -124.5

Figure 3.10 Finite element model used for gear system supported by bearings
67

Figure 3.11 Deformation diagram of bending in the axial direction for FEM (top) and KISSsoft (bottom)
68

3.1.3.4 Finite element model of housing

The housing consisted of two casing enfold the gear drive system illustrated in figure 3.12.
The outer bearing rings of all shafts are aligned with bearing seats and fixed to the housing
at specified bearing positions. The bonded contact excludes penetration, separation, and slid-
ing between bearings and housing faces or edges. The housing casing is simplified by re-
moving small fillets, holes, and features that have a negligible impact on the thesis investi-
gation. Casing finite element model includes the linear three-dimensional continuum of four
nodes tetrahedron elements. The total number of elements generated for both housing cas-
ings is approximately elements and the number of nodes is nodes.

Figure 3.12 The full finite element model of the helical stage. The upper part of the figure illustrates the outer
view on the model, while the bottom image shows components assembled inside the housing.
69

3.1.4 Finite element contact

This section introduced selected parameters for teeth contact behavior modeled utilizing fi-
nite elements. The gear teeth contact physics is non-linear and requires significant computa-
tional resources to solve. It is essential to consider the gear pair kinematics discussed in
Chapter 2.

For three-dimensional teeth contact in the mesh, the surface-to-surface contact model has
been selected, as interaction happens at relative unknown surface points on complex involute
formed surface. The surface-to-surface contact model supports large deformation, sliding,
and does not require compatible mesh between contact surfaces, which makes it suitable for
complex curved surface interactions. This model principle is to track the coordinates of
selected nodes and points on the contact surface concerning points on the surface area of the
target elements. The contact schematic is illustrated in figure 3.13.

Figure 3.13 Point to surface contact formulation schematic [79]

The penetration gap is estimated for each point on a contact body by searching the point
on the element of target body defined by the smallest Euclidean distance of the vector
between surface points. The contact problem represents the system of equation gen-
erated by exploiting the orthogonality condition between the vector and the line pass-
ing through point and tangent to the element surface. The tangent line is directed along
the vector of variation of each coordinate and . These coordinate parameters are obtained
from the Newton-Raphson iterative scheme. Then the minimal distance between contact-
ing elements is equal to the normal projection of the vector to the master element.
Finally, the penalty contact force is active when the gap between bodies becomes zero. Initial
penetration is obligatory to maintain equilibrium, while physically, there is no penetration.
Otherwise, the boundaries are not satisfied, and rigid body error stops the solution procedure.

High-speed shaft teeth are the target bodies, and the corresponding teeth on the driving wheel
are contact bodies. Selected Gauss Point detection method increases contact detection reso-
lution as it uses more points on the contact element surface.
70

The Augmented Lagrange formulation method precisely forms the contact stiffness and re-
strain model from over-penetration. The pressure and friction stresses are tracked and aug-
mented for each equilibrium iteration. It allows to reduce penetration to smaller values than
allowable tolerance and tends to better conditioning.

The asymmetric formulation option allows estimating the surface-to-surface model effi-
ciently, so all contact elements are always located on one surface and all target elements on
the other. It also permits obtaining the actual contact pressure that is the average from both
contact surfaces.

The parametric models of both gears are connected in be tangent and generate initial pene-
tration to model contact accurately and avoid rigid body movement convergence error. "Add
offset 0 mm, no ramping" interface treatment option was selected to exclude any automatic
correction of initial contact position and perform contact analysis according to the algorithm
described above. According to the surface-to-surface contact model discussed above, initial
penetration is usually defined manually, the specific amount of loading substeps normalize
contact and reduce initial penetration value that makes contact analysis to be physically ap-
propriate, as illustrated in figure 3.14. Normal contact stiffness is updated during each iter-
ation to accomplish normalization as it defines the amount of penetration between contact
and target surfaces. It allows the penetration to stay acceptably small, together with an ap-
propriate rate of convergence.
71

Figure 3.14 Influence of initial penetration on contact physics, the upside is an initial penetration to obtain
convergence, and the bottom part is normalized penetration to obtain convergence.

As Hertz's contact theory does not consider the friction effect, the frictionless contact type
is used to make a comparison of both methods under the same Hertz hypothesis assumptions.
72

3.1.5 Solver procedure

Boundary conditions accomplish the power transmission path. The solution is obtained using
ANSYS Transient solver. Transient structural allows analyzing efficiently the response of a
structure under the action of any time-dependent loads, where the sources of structural non-
linearity presented. ANSYS Static Structural was evaluated but found inappropriate due to
a lack of available boundary tools and a lower rate of convergence in comparison with AN-
SYS Transient.

The quasi-static approach is implemented to estimate static transmission error. Components


deform statically, and the inertia effect is negligible because of constant loading, high contact
ratio, and absence of gear backlash. The time integration option is switched off due to the
static purposes of simulation. The large deformation option is used to allow gear motion.

The simulation time was limited to one mesh cycle period to satisfy computational limita-
tions and costs. The total solution time consists of specific substeps to perform a Fast Fourier
Transform analysis based on the appropriate sampling frequency and the required transmis-
sion error curve smoothness. Test simulations were made to identify the time step for precise
results. In terms of the current thesis project, the sampling frequency is equal to 1/10 of
harmonics frequency, and a time step is t=0.000014 s. According to the Fast Fourier Trans-
form theory, the sampling number is n=1024 and period T=0.0144 s. Full driving shaft rota-
tion requires high computational cost and is not necessary for static transmission error cases,
where the displacement is the same for each mesh cycle. The required amount of substeps
to depict one mesh cycle data is the optimum solution for satisfying the computational cost
goal. The ADPL algorithm presented in Appendix 2 removes unnecessary information from
the solver file and output only nodal data in a specified location. The data of nodal points
positioned at the center of gear wheel and high-speed shaft rotation is recorded to analyze
gear excitation response in terms of the transmission error and the time-dependent mesh
stiffness. The ADPL algorithm generates and exports the text file as time series with angular
rotational data of reference nodes in radians. This procedure delivers significant improve-
ment in required disk capacity for a resultant file from approximately 750 gigabytes to 5
megabytes.

3.1.6 Transmission error post-processing

The magnitude of the reference nodal rotational displacement and their respective point in
time illustrate response in terms of the static transmission error and variable mesh stiffness.
A common way to assess the dynamic response in the transmission is to utilized displace-
ment as a function of frequency instead of time implemented into the Python-based algo-
rithm presented in Appendix 1.

At first, the rotational displacement and time data for each node are imported. The transmis-
sion error and corresponding variable mesh stiffness are then approximated using equations
2.10 and 2.11. Where is the nodal angular displacement of driving gear and corresponding
driven angular displacement - . Second, the single mesh period is extracted from the trans-
mission error data and substitute into the new array to obtain required parameters of the
period and sample number for desired Fast Fourier Transform algorithm, which output the
magnitude of response of the examined static transmission error as a function of frequency.
73

3.2 Design of evaluation procedure

This section identifies the assessment criteria that demonstrate the achievement of applica-
bility outcomes of the selected contact analysis method. Also, this section presents the initial
conditions for comparison analysis between FEM and KISSsoft approaches based on assess-
ment criteria.

3.2.1 Assessment criteria

The appropriate method for tonality assessment is chosen concerning specific selection cri-
teria based on theoretical aspects and influence factors described in Chapter 2.

In the robust design of a transmission gearbox, the primary purpose is to develop a solution
of micro and macro geometry of elements to compromise load-caring capacity and accepta-
ble tonality behavior. The first requirement is the ability to perform an assessment of a large
number of gearbox configurations and match the realistic behavior of the transmission. The
second requirement is the ability to consider changes in the shaft, gear wheel, and housing
geometry that affects the stiffness of elements while running the strength analysis to keep
the safety of gear operation above planned values. The third requirement is the ability to
estimate tooth deflection while considering the influence of tooth profile modifications, mis-
alignments, and bearings configuration. These critical parameters affect the shaft deflections
and the gear load distributions that have an impact on the reliability of the system.

Accuracy of assessment is essential for the selection of potential geometry variable that pro-
vides criterion value for the peak to peak transmission error and becomes an indicator for
the robustness of a chosen design.
74

3.2.2 Comparison approach

The general approach for comparison in terms of assessment criteria provided in Section
3.3.2 is as follows:

1. Cases' goal is to study components influence on the resultant transmission error eval-
uated using both approaches to identify the range of applicability of the analytical
method implemented in KISSsoft AG software in comparison with the finite element
method. All cases are linked with each other as they model the same power transmis-
sion path with the same loading conditions. The first case identifies only tooth effect
on transmission error provided by both methods. The second case introduces inner
geometry and shafts to compare how both methods account for misalignments. The
third and fourth cases introduce support elements into the solution and aimed to iden-
tify correlations between observed results.

2. The helix correction angle of teeth was estimated and modified for each case to ex-
clude mesh gap, to balance inclined contact lines through the whole contact trace,
and to reduce an undesired load intensity experienced by compensation for the tor-
sional deflection of other components. The amount of modification of crowning and
tip relief is identical for all developed teeth models. All cases deliver the identical
function of unloaded transmission error.

3. The finite element model and analytical solution boundaries are established, corre-
spond to each other, and based on the desired power transmission path: selected cases
and boundary conditions discussed in Section 3.3.3.

4. Finite element analysis and analytical calculations are performed for each case under
the same initial conditions and compared in terms of transmission error, mesh stiff-
ness, contact stress states.

5. The influence of each method on derived results of the transmission error function
and frequency spectrum response is depicted.

6. Finally, the contribution of main gear design and supporting factors on mesh stiffness
variation are identified.

Results based on the mentioned approach are analyzed in Chapter 4.


75

3.2.3 Studied cases

This section gives the details of the contact analysis performed with the utilization of both
methods and initial conditions used to obtain the results discussed in Chapter 4.

The nominal loading conditions are used for each case - nominal torque (27187.8 N*m) was
applied to the pinion, and rotational velocity load (372.6 rpm) was defined for the driving
wheel. All cases support constrains of both methods were designed to accomplish the power
transmission path illustrated in figure 3.33 and satisfy gearing law. Correspondence in the
power transmission path allows the introduction of new components, identification of their
influence, definition of correlations between selected methods, and specification of the
source of differences.

Figure 3.15 Schematic diagram of the power transmission path on a helical gear system
76

3.2.3.1 Case 1

Figure 3.16 Boundary conditions for case 1. A torque is applied to the driving wheel cylindrical surface. The
inner cylindrical surfaces are constrained, so both gears have no rigid body rotation, and the actual driving
wheel rotation is calculated.

The objective of the first case is to study tooth compliance and consider tooth stiffness as
the primary influence element. In this case, the flexibility of other elements of the gearbox
system is negligible. Manufacturer specifications define the positioning, geometry dimen-
sions, and loading values. The simplified bodies represent the gear wheel and the high-speed
shaft. The revolute joints are applied to the mounting surface of both gears to fix them in a
space and allow rotation in the operational direction, as illustrated in figure 3.16. The gears
rotate around their centers, assure gearing law and power transmission along the line of ac-
tion, which is the common tangent to the base circles.
77

Boundaries conditions accomplish rotational velocities for the independent degree of free-
dom of driving wheel revolute joint to estimate actual driving wheel rotation coordinates,
and allow generating torque distributed along the output shaft to model the state of the power
transmission during the operational mode. The moment load is assigned to the rotational
degree of freedom in the operational direction of the driving wheel.

Shaft deflection and support factor calculations were disconnected in the analytical algo-
rithm to obtain the same kinematics during contact analysis according to the Hertzian
method.

3.2.3.2 Case 2

Figure 3.17 Boundary conditions and initial conditions for case 2 considering actual geometry of gear wheel
and shaft components
78

The goal second case is to study the effect of gear body geometry as well as tooth compliance
on the provided results. The shafts introduce system compliance that leads to misalignment.
The actual geometry of gear bodies and shafts is implemented. The small chamfers and fillets
feature that do not influence on rim thickness is neglected to reduce computational cost. The
revolute joints are applied to the actual mounting surface on the ring from the input side of
the driving wheel and the generator side mounting surface at the high-speed shaft to fix them
in a space and allow rotation in the operational direction as illustrated in figure 3.17. Bound-
aries' conditions are the same as in case 1. The reaction torque was assigned to the rotational
degree of freedom in operational direction acting from the generator mounting surface.

The shaft calculations were included in an analytical solution to obtain the same kinematics
during contact analysis according to the Hertzian method.

3.2.3.3 Case 3

Figure 3.18 Boundary conditions and initial conditions for case 3.


79

The objective of the third case is to study the effect of bearing supports and shafts as well as
tooth compliance. The model is fixed in space by applying fixed support constrains on bear-
ing outer ring. The gear kinematics are defined by remote rotational displacement constraints
to define movements, power transmission input and to exclude any reaction forces that do
not correspond to the actual drive system. Remote displacement is applied to the actual
mounting surface on the ring from the input side of the driving wheel to assign rotation in
the operational direction, as illustrated in figure 3.18. The axial force is assigned to model
external load influence from the main assembly line. The torque is applied from the generator
side, as in case 2.

The shaft and bearing calculations were included in an analytical solution to obtain the same
kinematics during contact analysis according to Hertzian method
80

3.2.3.4 Case 4

Figure 3.19 Boundary conditions for case 4.

Finally, the fourth case investigates the influence of housing supports. The complete gearbox
stage system is used. The model is fixed in the space by applying fixed support constrains
on the outer housing surface, as illustrated in figure 3.19. Remote displacement and applied
torque are assigned in the same way as in case 3.
81

4
First of all, the resulting time-varying transmission errors obtained by FEM are compared
with the analytical solution. Second, demonstrate criterion value for the
peak to peak transmission error under assessment criteria. Third, the expected tonality and
contribution of transmission components on transmission error function are evaluated.
Fourth, transmission components influence the gear load distributions. The corresponding
impact on the reliability of the system is assessed. Fives, the stress response of the gears in
the mesh is compared against the analytical solution. And finally, the bearing modeling issue
is introduced. The conclusions are presented in Chapter 5.

4.1 Calculated Transmission Error and

This section delivers results of elastic contact analysis between the mating gear teeth and
accounts contribution of different components on resulted transmission error function. The
criterion value for the peak to peak transmission error is found and discussed.

4.1.1 Results of case 1

The first case transmission error and corresponding system mesh stiffness results from both
methods are illustrated in figure 4.1 and figure 4.2 for one mesh cycle at the time domain.
The static transmission error and secant mesh stiffness function are closed to the observa-
tions made by most research publications [15] [29] [1], which indicates the systematic rota-
tion through the mesh cycle of the gear teeth pair. The secant stiffness and transmission error
function domain slightly differ as KISSsoft provides a bit harder stiffness value. The initial
penetration requirement for finite element contact formulation may be a reason for this dif-
ference.

The local maximum and minimum peak coordinates correspond to handover points, as de-
picted in figure 4.2, where the transition between the load-carrying number of teeth pairs in
contact occurs. It affects the change of system stiffness estimated with equation 2.11.
FEM is suitable for visualization of the transition mechanics presented in figure 4.2. The top
point 1 in figure 4.3 corresponds to HPSTC point, and the bottom point 4 corresponds to
LPSTC for both cases where the significant change of mesh stiffness occurs. Contact pro-
gresses from the root to the tip of driving gear and opposite on the driven gear. HPSTC and
LPSTC points are criterion variables. The transition between these points is the peak of peak
value that forms the first harmonics illustrated in figure 4.4 and has a significant influence
on the internal dynamics of excitation. The total peak to peak value provided by FEM is 1.3
and by KISSsoft is 1.7 . Change in shared contact between the gear teeth pairs forms
the bottom crown at points 2-3-4 in figure 4.3 since contact between individual teeth pairs
is happening in mid-region where crowning, tip, and root relief affect the contact pattern
across teeth face. These transitions are responsible for growth in the second and third har-
monics, illustrated in figure 4.4, which amplitude is approximately the same. The corre-
sponding sound pressure level, according to Masuda [33], is 81.9 [dB(A)] provided by
KISSsoft and 78.1 [dB(A)] provided by FEM. However, the FEM approach provides a
smooth function in comparison with the KISSsoft method. Linear relief is in contact through
the whole pattern and results in small stepped fluctuations. This effect is discussed in Section
2.8 and observed in all analytical results.
82

Figure 4.1 Time variant transmission error curve for one mesh period of case 1 set up for FE-based solution
and KISSsoft

Figure 4.2 Time variant secant mesh stiffness curve for one mesh period of case 1 set up, where C is pitch
point, D - HPSTC, B - LPSTC.
83

Figure 4.3 FEM contact pattern of gears at local maximum and minimum of gear mesh stiffness provided by
FEM for case 1
84

Figure 4.4 Frequency spectrum of case 1 for (a) Finite element based solution and (b) KISSsoft

4.1.2 Results of case 2

The second case transmission error from both methods is illustrated in figure 4.5 for one
mesh cycle at the time domain. The observations made for case 1 are valid for the results of
case 2. The local maximum and minimum peak coordinates correspond to handover points,
where the number of teeth in contact variate and change corresponding system stiffness ,
as depicted in figure 4.6. The transition mechanics between local maximum and minimum
points are illustrated in figure 4.7. According to these contact, pattern results from case 1 in
figure 4.3 and case 2 in figure 4.7, the tooth profile modifications cause first local minimum
at t=0,0078 s illustrated in figure 4.6 as most of the contact occurs on the modified area.

However, the difference between the provided results decreases, when shaft deflection is
considered. The total peak to peak value provided by FEM is 1.4 , and KISSsoft is 1.3
Sound pressure level, according to Masuda [33], is 78.1 [dB(A)] provided by KISSsoft
and 78.6 [dB(A)] provided by FEM.

The obtained results of transmission error deliver a small difference in the function shape
caused by a difference in the formulation of shared contact between teeth by FEM and
KISSsoft. Gear and shaft geometries influence torsional deflection and cause changes in the
contact pattern that explain the difference in the shape of obtained results. As stated previ-
ously, KISSsoft does not consider the effect of the actual geometry of gear and shaft, which
results in disagreement with spectrum results presented in figure 4.7.
85

Figure 4.5 Time variant transmission error curve for one mesh period of case 2 set up for FE based solution
and KISSsoft

Figure 4.6 Time variant secant mesh stiffness curve for one mesh period of case 2 set up, where C is pitch
point, D - HPSTC, B - LPSTC.
86

Figure 4.7 FEM contact pattern of gears at local maximum and minimum of gear mesh stiffness for case 2

Figure 4.8 Frequency spectrum of case 2 for (a) Finite element based solution and (b) KISSsoft
87

4.1.3 Results of case 3

The third case transmission error from both methods is illustrated in figure 4.9 for one mesh
cycle at the time domain. Results display a significant difference in shape, system stiffness,
and corresponding spectrum.

First of all, the transmission error domain of KISSsoft results (-72 to -74 ) significantly
differs from FEM results ( -169 to -171 ), as illustrated in figure 4.9. It can be explained
by bearing performance and its effect on the gear drive system discussed in section 4.3.
These bearing affect the shaft deflections and the gear load distributions that have an impact
on the output mesh system function, as illustrated in figure 4.10. It is also illustrated in figure
3.11 that the total shape of high-speed shaft deflection differs from analytical results. Ini-
tially, profile modifications participated in contact at time instance in figure 4.10
lead in reduction of stiffness and generation of the local minimum. FE-based bearing perfor-
mance changes the shared contact pattern illustrated in figure 4.11, resulting in a sinusoidal
transmission error function where the LPSTC point of contact path moves from point 2 to
point 1.

The total peak to peak value provided by FEM is 2.0 , and by KISSsoft is 1.3 This
difference affects the spectrum results illustrated in figure 4.12. Sound pressure level, ac-
cording to Masuda [33], is 78.1 provided by KISSsoft and 82 provided by
FEM.

Figure 4.9 Time variant transmission error curve for one mesh period of case 3 set up for FE-based solution
and KISSsoft
88

Figure 4.10 Time variant secant mesh stiffness curve for one mesh period of case 3 set up, where C is pitch
point, D - HPSTC, B - LPSTC.

Figure 4.11 FEM contact pattern of gears at local maximum and minimum of gear mesh stiffness for case 3
89

Figure 4.12 Frequency spectrum of case 3 for (a) Finite element based solution and (b) KISSsoft

4.1.4 Results of case 4

The last case transmission error results from both methods are illustrated in figure 4.13 for
one mesh cycle at the time domain. The observations made for case 3 are valid for results of
case 4 illustrated in figures 4.13-4.16. The housing casing causes a minor reduction in system
mesh stiffness illustrated in figure 4.14. The total peak to peak value provided by FEM is
1.9 , . The sound pressure level, according to Masuda [33], is
78 [dB(A)] provided by KISSsoft and 82 [dB(A)] provided by FEM.

Figure 4.13 Time variant transmission error curve for one mesh period of case 4 set up for FE-based solution
and KISSsoft
90

Figure 4.14 Time variant secant mesh stiffness curve for one mesh period of case 4 set up for Finite element
based solution and KISSsoft

Figure 4.15 FEM Contact pattern of gears at local maximum and minimum of gear mesh stiffness for case 4
91

Figure 4.16 Frequency spectrum of case 4 for a) Finite element based solution and b) KISSsoft
92

4.2 Stress states

Due to the visualization possibilities of both tools for the contact stress results, the scalar
value of the contact stress is detected at the main points of contact path, and these values can
be found in table 4.1 for Hertzian pressure provided by KISSsoft and contact pressure field
from ANSYS Workbench.

Both results correspond to the observation that contact stress decreases when a line of action
increased. It is possible to observe that contact deformation in terms of transmission error
agrees well; meanwhile, Ansys deliver lower maximum contact stress then KISSsoft. FE-
model has more nodes to detect contact than in the KISSsoft analysis. Instant variation of
the normal force application angle occurs between a pure involute profile and profile modi-
fication and causes stress peaks detected on the tooth profile of KISSsoft results. This ob-
servation is described and explained in section 2.8. Figure 4.17 illustrates that stress values
significantly increase close to the tip and root of the teeth due to applied crowning and reliefs.
This observation is valid for all cases. More nodes and elements in the FEM model lead to
smoother stress distribution, so the maximum value is less than with KISSsoft.

Table 4.1 Maximum contact pressure at the contact path provided by KISSsoft method and FEM method

[MPa] PP [MPa] [MPa]


KISSsoft
Case 1 1169 1101 1028
Case 2 1134 1061 980
Case 3 1123 1063 988
Case 4 1125 1065 992
ANSYS
Case 1 938 931 920
Case 2 936 927 913
Case 3 971 966 952
Case 4 973 967 956
93

Figure 4.17 Contact stress distribution of case 1 along the face width for one tooth during one mesh cycle
provided by KISSsoft

The root stress results are detected at the main points of contact path on both gears, and these
values can be found in table 4.2, and table 4.3 for tooth root stress provided by KISSsoft and
maximum principal stress field approximated in ANSYS Workbench. The highest stress is
observed on HPSTC point where the lowest number of teeth participate in mesh.

The KISSsoft uses virtual spur gear pair to approximate helical gears. According to KISSsoft
documentation [5], KISSsoft uses standard-based formulas for tooth root formulas that valid
for a standard involute profile, and that is not true for the studied cases. FEM accounts geo-
metrical effect on tooth root stress, that justify significant variation in observed result. Due
to this advantage, FEM precisely investigates whether the maximum root stress occurs.
94

Table 4.2 Maximum tooth root stress on the high-speed shaft during the significant time instances of the contact
path provided by ISO 6336 and FEM method

[MPa] PP [MPa] [MPa]


KISSsoft
Case 1 172 187 217
Case 2 163 179 208
Case 3 160 178 206
Case 4 163 178 207
ANSYS
Case 1 216 248 273
Case 2 211 236 266
Case 3 282 272 293
Case 4 283 271 293

Table 4.3 Maximum tooth root stress on driving wheel during the major time instances of contact path provided
by ISO 6336 and FEM method

[MPa] PP [MPa] [MPa]


KISSsoft
Case 1 201 205 179
Case 2 190 195 172
Case 3 189 193 171
Case 4 190 195 173
ANSYS
Case 1 278 270 269
Case 2 276 269 264
Case 3 302 298 290
Case 4 304 297 290
95

4.3 Bearing excitations

Following the gear noise theory, the bearing fluctuation should be assessed. The excitation
force originated from transmission error and rotary fluctuation of shaft components induce
bearing excitation forces. The third case variation of bearing reaction force on the high-speed
shaft is studied, and results from both methods are illustrated in figure 4.18 and figure 4.19
for one mesh cycle at the time domain. T1 and T2 are the bearings installed on high-speed
shaft and illustrated in figure 3.10.

The interaction of transmission error and several mechanical effects from related compo-
nents generate bearing reaction forces fluctuations that make them involved in the assess-
ment of vibration. The detailed parameters of the bearing are essential to establish accurate
analysis. The simplified lumped bearing model is implemented to reduce computational cost.
Bushing joint does not allow to use of asymmetric stiffness matrix, which causes divergence
since KISSsoft and BEARINX works with asymmetric stiffness matrixes. Other options are
not available or suitable for modeling of complex bearing joints. Ansys joints do not allow
to assign a stiffness matrix to any connections except bushing.

FEM excitations result partly represents a similar behavior in shape but leads to reduced
amplitudes. Initial reaction forces provided by standard calculation and FEM justify bearing
modeling way for static analysis, but the corresponding excitation amplitudes display non-
negligible differences that explain the inadequate representation of bearing performance dur-
ing the contact analysis. However, it is impossible to make conclusions without validation
tests.

The excitation of bearings located on the input shaft is not estimated by KISSsoft and cannot
be assessed from contact analysis.
96

Figure 4.18 Excitation of total bearing reaction forces for single mesh period of the case 3 bearing T1 provided
by KISSsoft and FE-based solution.

Figure 4.19 Excitation of total bearing reaction forces for single mesh period of the case 3 bearing T2 provided
by KISSsoft and FE-based solution.
97

5
5.1 Conclusions

The work is accomplished in the field of gear research to identify a range of applicability of
the analytical method proposed by Weber und Banaschek [3] and implemented in KISSsoft
AG software in comparison with the finite element method and recommend virtual testing
of transmission error for conceptual wind gearboxes design use.

Different study cases were carried out using the KISSsoft gear calculation tool and ANSYS
Workbench. The thesis studies mainly revolved around the theory of transmission error. Ac-
cording to the concept of transmission error theory, results of contact analysis define crite-
rion parameter for one of the leading causes of gear noise. Criterion parameters are primary
points of interest during the optimization approach to tonality free gearing. The work also
investigated the causes of the observed differences between solutions provided by KISSsoft
and FEM. Both results provide meaningful correlation and dissimilarities between KISSsoft
and FEM approach in terms of affecting factors discussed below. The following conclusions
are made:

The findings from both approaches depict that there is a strong correlation between
transmission error profile results from the FEM and KISSsoft. The peak-to-peak STE
predicted by FEM is close to that obtained using KISSsoft.

The peak to peak amplitude is formed by corresponding local minimum and maxi-
mum mesh stiffness values at LPSTC and HPSTC coordinates of the contact pattern.
It is the criterion value of noise generation.

This peak to peak value is used to approximate generated noise pressure level fol-
lowing empirical equation 2.12 that has been proposed by Masuda [33]. The differ-
ence between predicted peak to peak values is not significant in terms of generated
noise pressure level.

According to noise theory, the main affecting factor on gear whine is tonality ampli-
tude, which difference is significant case by case. The accuracy of amplitude meas-
urements is not validated.

The location of the contact pattern within the parametric teeth model of the analyzed
gear pair represents a significant factor in predicting the outcome of the transmission
error function.

The parameterized variables affecting transmission error function variation and cor-
responding spectrum are quantified as a function of tooth profile modification and
contact ratio.

The accuracy of elements modeling affects the shaft deflections and corresponding
gear load distributions that have an impact on the reliability of results.
98

The KISSsoft gear calculation tool overestimates peak to peak transmission error,
contact stress states and underestimates tooth root stress in comparison with FEM.

KISSsoft ignores gear body geometry, gear rim thickness, cross-influence. However,
the parametric tooth model, gear inner geometry, and shaft geometry provided by
Moventas do not significantly change the shape of the transmission error curve pro-
vided by FEM in comparison with the KISSsoft analytical models.

Both solutions account for misalignments caused by housing deflection, which do


not lead to significant changes in results.

KISSsoft method based on experimental researches held on standard gear sets [48]
that include assumptions discussed in Section 3.3, so it does not accurately approxi-
mate stress states on helical gears with high contact ratio and presented microgeom-
etry modifications during contract performance.

FEM approach requisite high computational cost in comparison with the analytical
approach, as it requires extremely high resolution across the entire path of contact.
Gear body geometry, gear rim thickness, and cross-influence are modeled com-
pletely.

Helical gear teeth contact is modeled directly but requires small initial penetration
that may affect system mesh stiffness value.

The selected bearing approximation method in ANSYS provides an inadequate rep-


resentation of bearing performance during contact analysis.

The finite element method should provide realistic results and have the potential to become
an essential tool in tonality free optimization studies. However, production time plays a sig-
nificant role in the wind power gear industry. Engineers must quickly modify the macro and
micro geometry of gears case by case. It is not feasible to model each concept design using
the finite element approach as it requires enormous computational cost. KISSsoft is a suita-
ble tool and delivers the required range of applicability to find the optimized shape that de-
livers the lowest STE amplitude. KISSsoft is able to estimate tooth deflection while consid-
ering the influence of tooth profile modifications, misalignments caused by system configu-
rations. Then the final version is evaluated by the FEM tool, where the complex powertrain
transfer behavior and near-field sound pressure distribution is assessed from generated bear-
ing reaction force fluctuations.

The accurate representation of bearing supports and corresponding excitation are the most
critical for the further modal and acoustic studies. KISSsoft cannot provide accurate acoustic
and modal analysis results, as it is impossible to include realistic housing geometry into the
script. Thus, FEM should be used to evaluate the tonality of the gearbox.

Finally, both tools can be used together in optimization studies of powertrain by defining the
optimum combination of macro and micro gear modifications to minimize tonality ampli-
tudes and corresponding gear casing geometry to reduce generated whine noise to a satisfac-
tory level.
99

5.2 Recommendation for future work

The main recommendations are based on the development of the current FEM approach. The
current numerical model should be developed to accurately represent bearing supports to
achieve a realistic representation of the stage performance.

Experimental measurements should be made to confirm predictions made in this thesis work,
to validate the finite element model and make conclusions about the accuracy of both meth-
ods.

Other studies should analyze the effects of load and design variables on transmission error.
The meta-model can be built based on obtained results to use data science tools to predict
and optimized design solution for multiply objective functions, that include transmission
error and stress states in a mesh. This concept requires an enormous computational cost to
set up the model but may deliver benefits in long term optimization project.

The scope of these studies was limited only to the stage that includes a parallel pair of helical
gears. A new study should be initiated for planetary gear set and full gearbox assembly,
where the effect of variable loading and cross-influence from transmission error of one stage
to another on generated noise should be assessed. These studies require significant compu-
tational and time resources, so the utilization of high-performance computers should be con-
sidered.
100

Bibliography

[1] J. Smith, Gear Noise and Vibration, New York: Marcel Dekker,Inc., 2003.
[2] A.Kahraman and
ASME Journal of Mechanical Design, pp. 112-118, 1999.
[3] - und
Schriftenreihe Antriebstechnik, Heft 11, 1953.
[4]
Bubikon, Switzerland, 2019.
[5] KISSsoft
https://old.kisssoft.ag/Manual/en/index.htm. [23 March 2020].
[6] R. S. Pavlovich, Dudley's handbook of practical gear design and manufacture, Boca
Raton, FL33487-2742: Taylor&Francis Group, 2016.
[7] J. Shigley, C. Mischke and R. Budynas, Mechanical engineering design, 7th Edition,
New York, London: McGraw-Hill, 2003.
[8] B. Hamrock, B.Andcobson and S.Schmid, Fundamentals of Machine Elements,
London: McGrawHill, 1999.
[9] Ameri
-55589-846-7. OCLC 65562739, ANSI/AGMA 1012-G05,
2005.
[10] J. Smith, Gears and theirs vibration, A Basic Approach to Understanding Gear Noise,
New York: The Macmillan Press LTD, 1983.
[11] ; BGA Knowledge Transfer Programme, Nottingham, 2018.
[12]
The First Institution of Mechanical Engineers International Conference on Gearbox
Noise and Vibration, Cambridge, 1990.
[13] W.D.Mark, Performance-based gear metrology: kinematic-transmission-error
computation and diagnostic, John Willey&Sons Ltd, 2013.
[14] E.Hiroaki and
Prince Mohammad Bin Fahd University (PMU), Mechanical Engineering Department,
AlKhobar, 2012.
[15] D. R. Houser, F. B. Oswald, M. J. Valco, R. J. Drago and
of transmission error predictions with noise measurements for several spur and helical
In Proceedings of 30th AIAA, SAE, and ASEE Propulsion Conference,
Indianapolis, 1994.
[16] R. Gregory, S. Harris and
Journal of Scientific Instruments, volume/issue. vol 40, pp.
5-9, 1963.
[17] S.Wu and
Sound Vib. 317, pp. 608-624, 2008.
[18] L.Chang, G. Liu and
Mech. Mach. Theory, volume/issue. 46, pp. 1869-1887, 2011.
101

[19] R.G.Parker, S. Viandyakar and -linear dynamic response of a spur gear


Sound Vib. 237 (3), pp. 435-455,
2006.
[20] V. Ambarisha and
Sound Vib., volume/issue. 302 (3), pp. 577-599, 2007.
[21] N. Raghuwanshi and
photoelasticity Measurement, volume/issue. 73, pp. 439-452, 2015.
[22] C.Cooley, C.Lia, X.Dai and
Mechanism and Machine Theory, volume/issue. 105, pp. 540-
533, 2016.
[23] W. M
Journal of the Acoustical Society of America, 1978.
[24] T.Lin and
Mechanical
Systems and Signal Processing, volume/issue. 91, pp. 167-182, 2017.
[25] The dynamics of Parallel Axis Gear in an Automotive

[26] A.J.Lemanski, Gear Design, Warrendale: SAE, 1990.


[27] Y. Miura and
European conference on vehicle noise and vibration, pp. 1356-1448, 1998.
[28] J. C. Huang and -Backlash
Proceedings of the 1999 Noise and Vibration Conference, SAE paper,
1999.
[29] R. G. Munro and proceedings: The Gear
Noise Short Course, 2003.
[30] E. Nieuwenhuizen and s for wind turbines in
proceedings: EuroNoise 2015, Maastricht, 2015.
[31] S. Curtis, J. Pears, D. Palmer, M. Eccles, A. Poon, M. Kim, G. Jeon, J. Kim and S. Joo,

SAE Technical Paper 01 (1819), pp. 1-10, 2005.


[32] l knowleadge of gear noise - In IMechE
Conference on Noise and Vibrations of Engines and Transmissions 1979, Conference
Publication 1979-10, paper C117/79, pp. 9-14, 1979.
[33] T. Masudo, T. Abe and e Considering the
Journal of Vibration, Acoustics, Stress
and reability in Design, volume/issue. Vol. 108/95, 1986.
[34] M.Iglesias, A.Fernández, A.Juan, A.Díez, P.García and
Eccentricity Er In:
Pennacchi P. (eds) Proceedings of the 9th IFToMM International Conference on Rotor
Dynamics. Mechanisms and Machine Science. Springer, Cham, volume/issue. vol 21,
pp. 1381-1390, 2015.
[35] H. Lin and
Memphis, Memphis, Tennessee, 1998.
[36] M. Umezawa, T. Suzuki and
Bulletin of JSME, pp. 397-402, 1958.
102

[37] W.Yu and


Mechanics and Machine Theory, volume/issue. 96, pp. 146-164, 2016.
[38] L. Zhao, M. Du and
International
Journal of Materials, Mechanics and Manufacturing, volume/issue. 6, pp. 1-4, 2018.
[39] J.Klingelnberg, Bevel gear: Fundamentals and applications., Springer, 2016.
[40] Proceedings of the Institution
of Mechanical Engineers, volume/issue. 172(1), pp. 87-112, 1958.
[41] Proceedings of the Institution
of Mechanical Engineering, volume/issue. 172(2), pp. 87-100, 1958.
[42] Y. Cai and
gears to m In ASME International Power
Transmission and Gearing Conference, volume/issue. 43, p. 453 460, 1992.
[43] S. Matsumura, K. Umezawa and
pair having tooth surface Proceedings of
Power Transmission and Gearing Conference ASME, volume/issue. 88, pp. 161-168,
1996.
[44]
Proceedings of Power Transmission and Gearing Conference ASME,
volume/issue. 88, pp. 17-21, 1996.
[45] R. Budynas and J. Keith, Shigley's mechanical engineering design McGraw-Hill series
in mechanical engineering, Singapore: McGraw-Hill, 2011.
[46] R. C. Juvinall and K. M. Marshek, Fundamentals of machine component design 5th
ed., Hoboken: John Wiley & Sons, 2012.
[47]
Gears with Any Fillet Geometry under Consideration of the Exact Meshing
Proceedings of the Institution of Mechanical Engineers, Part C: Journal
of Mechanical Engineering Science, volume/issue. 210, pp. 1989-1996, 2019.
[48] KISSsoft AG - -A
Gleason Company, [Online]. Available:
https://www.kisssoft.ch/Manual/en/10334.htm. [28 January 2020].
[49] - Proceedings of the Institution of
Mechanical Engineers, volume/issue. 163, pp. 162-175, 1950.
[50] R. Gregory, S. Harris and Applied
Mechanics Group, volume/issue. 178, pp. 207-221, 1962.
[51] H.Yang and ASME Journal of
Mechanisms, Transmissions and Automation in Design, volume/issue. 107, pp. 529-
535, 1985.
[52] A.Kahraman and - Journal of Sound
and Vibration, volume/issue. 142, nro 1, pp. 49-75, 1990.
[53] E. Budak and - Linear
Structures with Local Non-Linearities and the Study of the Validity of Some
proceedings: The Fourth International Conference on Recent
Advances in Structural Dynamics, Southampton, England, 1991.
103

[54] R. Singh and - Journal of


Sound, pp. 49-75, 1990.
[55] R. Singh, D. Houser and -linear dynamic analysis of geared
-C-020, 1990.
[56] -linear mathematical model for dynamic analysis of spur gears
Journal of Sound and Vibration , volume/issue.
145, nro 2, pp. 239-260, 1991.
[57] A.Kahraman and Between Time-Varying Mesh Stiffness and
Clearance Non- Journal of Sound and Vibration,
volume/issue. 146, pp. 135-156, 1991.
[58] M.Kang and
compliant s Mechanical Systems and Signal Processing, volume/issue. 29, pp.
391-403, 2012.
[59] H.Lin, F.Oswald and
Mechanism and Machine Theory,
volume/issue. 29, pp. 1115-1129, 1994.
[60] H.Özgüven, R.Maliha and -Shaft-
Transactions of
ASME, volume/issue. 126, pp. 534-541, 2004.
[61] J.Morgan, M.Dhulipudi, R.Yakoub and els for
Assessing Gear Rattle and Gear Whine of Torque Transmission Systems with
proceedings: Noise and Vibration Conference and Exhibition,
St. Charles, Illinois, 2007.
[62] Static and Dynamic Finite-

Brunel University, Brunel, United Kingdom, 2009.


[63] H.Ozguven, R.Maliha and -Shaft-
Bearing Systems Usin Transactions of
ASME, volume/issue. 126, pp. 534-541, 2004.
[64]
Excitation by Using Describing Functions and Finite Element Met
Technical University, Ankara, Türkiye , 1994.
[65] A.Singh and -
Society of Automotive Engineers, SAE 941761, 1994.
[66] I. M. Allison and
Experimental Mechanics, volume/issue. 20, pp. 217-225, 1980.
[67] R.Parker, S.Viandyakar and -linear dynamic response of a spur gear pair
- Journal of Sound and Vibration,
volume/issue. 237, nro 3, p. 435 455, 2000.
[68] S.Baud and cal Geared
Systems- Journal of Mechanical Design,
volume/issue. 124, nro 2, pp. 334-346, 2002.
[69] M.Ajmi and -static and dynamic behaviour
of solid wide-faced spur and helica Mechanism and Machine Theory,
volume/issue. 40, nro 2, pp. 173-190, 2005.
104

[70] E.Rigaud and


VDI Berichte, volume/issue. 1996, pp. 883-
845, 1996.
[71] J.Zhou, W.Sun and -Noise Design Method Based on Panel
Mathematical Problems in Engineering, pp. 1-10, 2014.
[72] Y.Guo, T.Eritenel, T.Ericson and -acoustic propagation of gear
dynamics in a gear-bearing- Journal of Sound and Vibration,
volume/issue. 333, nro 22, pp. 5762-5785, 2014.
[73] Journal für reine und
angewandte Mathematik, pp. 156-171, 1881.
[74] KISSsoft AG - - A Gleason
Company, [Online]. Available: https://www.kisssoft.ch/Manual/en/10863.htm.
[Haettu 30 01 2020].
[75] H. Winter and Journal
"Antriebstechnik", volume/issue. 22, nro 3, pp. 39-42, 1983.
[76] H. Winter and Drive
technology, volume/issue. 22, nro 3, pp. 51-58, 1983.
[77] -ISO 6336-1:en. Calculation of load capacity
of spur and helical gears -- Part 1: Basic principles, introduction and general influence
25/05/2012.[Online].Available:
https://sales.sfs.fi/en/index/tuotteet/SFS/ISO/ID2/6/186539.html.stx. [31 01 2020].
[78] Antriebstechnik, 2011.
[79] P.Wriggers, Computational Contact Mechanics, Springer, 2006.
[80] D. Dudley, Handbook of Practical Gear Design, McGraw-Hill Inc., 1984, 1984.
[81] R. Mobley, Vibration Fundamentals, Elsevier, 1999.
[82] -21:2016, Industrial
automation systems and integration Product data representation and exchange
Part 21: Implementation methods: Clear
2016. [Online]. Available: https://www.iso.org/standard/63141.html.[10 02 2020].
[83] A.Kahraman and - Journal of Sound
and Vibration, volume/issue. 172, pp. 49-75, 1990.
[84] A. Toda and
ASME paper, nro DET-73, 1979.
[85] C.Hasl, H.Liu, P.Oster, T.Tobie and
stress of plastic spur gears meshing with steel gears under consideration of deflection-
Mechanism and Machine Theory, volume/issue. 111, pp. 152-
163, 2017.
[86] D. Houser, F. Oswald, M. Valko, R. Drago and
Transmission Error Prediction with Noise Measurements for Several Spur and Helical
proceedings: 30th Joint Propulsion Conference, Indianapolis, 1994.
[87] J.He and Modal Analysis, pp. 140-
158, 2001.
[88] M. Friswell, J. Penny, S. Garvey and A. Lees, Dynamics of Rotating Machinery, New
York, NY 10013-2473, USA: Cambridge University Press, 2010.
105

[89] R. C. Jr and A. Kurdila, Fundamentals of Structural Dynamics, 2nd Edition, New


Jersey: John Wiley & Sons Inc, 2006.
[90] T. Conry and
of Load Distribution and Optimal Modifications for Gear Systems, Engineering for
Industry, volume/issue. Vol.94, pp. 1115-1121, 1972.

Attachments

Appendix 1. Post-processing Python code for STE, mesh stiffness, and spectrum analysis.
3 pages.

Appendix 2. ADPL code to reduce computational cost in terms of internal memory require-
ments for ANSYS solution. 2 pages

Appendix 3. Helical gear dimensions and parameters. 2 pages


1
2
3
4

The general code connected to analysis settings for monitoring requested node data
5

This code was utilized in Workbench result section to write node results into text output
file
6
7

You might also like