FULLTEXT01

Towards model-based development of heavy truck
synchronizers
DANIEL HÄGGSTRÖM
Doctoral Thesis
Stockholm, Sweden 2018
KTH School of Industrial Engineering and Management
TRITA-ITM-AVL 2018:51 SE-100 44 Stockholm
ISBN 978-91-7873-002-5 SWEDEN
Academic thesis, which with the approval of KTH Royal Institute of Technology,
will be presented for public review in fulfillment of the requirements for a Doctor
of Engineering in Machine Design. The public review is held in Gladan (B319),
Kungliga Tekniska Högskolan, Brinellvägen 85, 3rd floor, Stockholm on November
22, 2018 at 10:00. © Daniel Häggström, November 2018
Print: Universitetsservice US AB
iii
Abstract
Gear shifts are becoming more and more important as engines are adapted
to low speed and high torque working conditions. Synchronizers are key
components for successful gear shifts. To adapt the synchronizers to future
engines and thus new working conditions, improved development tools are
needed. This thesis includes a comprehensive gear shift and synchronizer
frame of reference section with detailed explanations of how a synchronizer
works. The thesis also presents two types of numerical models that enable
design analyses of the synchronization process.
Fluid-structure interaction models were used to simulate oil evacuation be-
tween the synchronizer cones to assess the synchronizer performance during
the pre-synchronization phase. For the main synchronization phase, thermo-
mechanical finite element models were used to simulate the transient tem-
perature in the synchronizer contact surfaces. To verify and validate the
thermomechanical simulations, both bulk and surface temperature measure-
ments were used, as well as a qualitative comparison of the position of initial
wear marks relative to the position of high surface temperature areas in the
simulation. The validated thermomechanical model was used to predict fail-
ure in molybdenum coated synchronizers. It was shown that the simulated
temperature is a better predictor of synchronizer failure than the commonly
used parameters “synchronization energy” and “synchronization power”.
A methodology to develop friction models based on sliding speed, contact
pressure and surface temperature was developed, and applied for a molybde-
num coated synchronizer.
To allow for improved accuracy of carbon fiber reinforced polymer (CFRP)
lined synchronizer simulations, material data for a CFRP friction lining was
estimated with different test methods. A contact surface temperature thresh-
old where a reduction in coefficient of friction, accelerated wear and formation
of hot spots starts was identified.
To reduce the maximum surface temperature, a relative angle between the
cone contact surfaces can be introduced. The optimum relative cone angle
was determined based on measured geometric deviations for a population of
manufactured synchronizers. It was shown that there is an optimum relative
cone angle that significantly can reduce the maximum surface temperature
during synchronization.
Keywords: synchronization, synchronizers, gear shift, gearbox

iv
Sammanfattning
Växellådans växlingsprestanda får allt större betydelse. Synkroniserings-

enheter är viktiga komponenter för att kunna åstadkomma snabba och till-
förlitliga växlingar. För att effektivt kunna anpassa synkroniseringsenheterna
till nya arbetsförhållanden och belastningar krävs nya utvecklingsverktyg.
Denna avhandling innehåller ett utförligt kapitel om synkroniserings- och
växlingsteori med en detaljerad beskrivning av hur synkroniseringsenheterna
fungerar. Avhandlingen beskriver även två typer av numeriska modeller, som
har utvecklats för att möjliggöra simuleringar av synkroniseringsförloppet.
För att simulera försynkroniseringsfasens prestanda, dvs. hur snabbt oljan
mellan synkroniseringsytorna trycks ut under försynkroniseringen, användes
kopplade flödes- och stelkroppsmekaniska modeller. För synkroniseringsfasen,
dvs. från den tidpunkt när synkroniseringsytorna kommer i kontakt, användes
termomekaniska finita elementmodeller för att simulera den transienta tempe-
raturfördelingen i kontaktytorna. För att verifiera och validera de termomeka-
niska simuleringarna används både bulk- och yttemperaturmätningar, samt
en visuell jämförelse av nötningsmärkens position på kontaktytorna och posi-
tionen för den högsta simulerade kontakttemperaturen. Den validerade termo-
mekaniska modellen användes för att prediktera haveri i molybdenbelagda
synkroniseringsenheter. Det visades att den högsta simulerade kontakttempe-
raturen under synkroniseringen med större noggrannhet kan användas för att
prediktera haveri än de traditionella parametrarna “synkroniseringsenergi”
och “synkroniseringseffekt”.
En ny modelleringsmetod för hur friktionen i en synkroniseringsenhet vari-
erar under synkroniseringsförloppet, baserat på glidhastighet, kontakttryck
och yttemperatur, har utvecklats.
För att ge ökad noggrannhet vid simulering av CFRP-belagda synkroni-
seringsenheter har materialdata för de tunna vävda lager av CFRP uppmätts
med olika provmetoder. Kontakttemperaturen då friktionstalet börjar minska,
nötningshastigheten ökar och värmefläckar, så kallade “hot spots”, börjar
utvecklas identifierades. För att minska den maximala kontakttemperaturen
under en synkronisering kan de båda interagerande konytorna tillverkas med
en liten vinkelskillnad. Den optimala relativa vinkeln, dvs. den som ger lägst
maximal kontakttemperatur, bestämdes utifrån en population av uppmätta
synkroniseringskomponenter.
Nyckelord: synkronisering, synkroniseringsenheter, växling, växellåda

Preface
The research presented in this thesis was carried out between January 2014 and July
2018 at Scania transmission development in Södertälje, Sweden and the Department of
Machine Design, KTH Royal Institute of Technology in Stockholm, Sweden. January 2014
through December 2016 was financed by Scania CV AB, Volvo Group Trucks Technology
and VINNOVA (project No. 2112-04619). January 2017 to the end of the project was
financed by Scania CV AB. I’m deeply grateful for all the financial support received, which
made this project possible.
A special thanks to my supervisors Ulf Sellgren and Stefan Björklund. I am grateful
for their support and for all our productive discussions. Their guidance has been crucial
for this work. Thank you Kenth Hellström, project manager (2014-2016) for this project
as well as my manager at Scania, who has undeniably played a big role in making this
project happen and keeping it running smoothly and timely. When deciding what project
out of all possible candidates to prioritize, I’m thankful I had Göran Sandgren and Erik
Sandqvist on my side. In 2017, Olle Lundvall took over as project manager, and Sara
Molneryd Stjerndahl as my manager at Scania. I would like to thank you for your interest
and support during this project.
Thanks to the project team in Göteborg, Irfan Muhammad, Viktor Berbyuk and Håkan
Johansson at Chalmers University of Technology and Magnus Andersson at Volvo Group
Trucks Technology. Thanks to my co-authors Wiktor Stenström and Pär Nyman, as well
as Karl Vestgöte, who performed the physical test in the µ-COMP rig. Thank you Niklas
Melin, Scania CV AB, for all help related to my simulations. Thanks to Dan Jönsson
and Mattias Forslund at Scania CV AB Materials Technology, for the testing and support
regarding CFRP. Thanks to Monica Norrby, KTH Royal Institute of Technology and Jacob
Steggo, Jönköping University, for the CFRP material tests. Thanks to Anders Hedman
at Volvo Group Trucks Technology and Peer Norberg, Gustav Göransson and Mårten
Dahlbäck at Scania CV AB as well as my brothers Johan and Martin for reviewing my
thesis and papers.
Thank you colleagues, both at Scania and KTH, for making the work interesting and
fun, and for teaching me new things all the time.
Last, but not least, I want to thank my family. My parents Lena and Bo for their
unconditional support. I certainly could not have done this without the support of my wife
Tova, and you taking most of the “VAB” has been appreciated. Our daughters Wilma and
Alice probably didn’t help at all, and all the long sleepless nights certainly didn’t improve
my efficiency, but at the end of the day she showed what matters most in life.
v
Contents
1 Introduction 1
1.1 Research objective and motivation . . . . . . . . . . . . . . . . . . . 2
1.2 Research questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Thesis outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Swedish transmission cluster . . . . . . . . . . . . . . . . . . . . . . . 5
2 Frame of Reference 7
2.1 Simulation-driven product development . . . . . . . . . . . . . . . . 7
2.2 Gear shift in heavy truck transmissions . . . . . . . . . . . . . . . . 9
2.3 Synchronizer background information . . . . . . . . . . . . . . . . . 11
2.3.1 Related work . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.2 Different phases of synchronization . . . . . . . . . . . . . . . 18
2.3.3 Preventing unsynchronized gear engagement . . . . . . . . . . 19
2.3.4 Pre-synchronization phase . . . . . . . . . . . . . . . . . . . . 21
2.3.5 Main synchronization phase . . . . . . . . . . . . . . . . . . . 22
2.3.6 Synchronizer wear and wear gap . . . . . . . . . . . . . . . . 23
2.3.7 Synchronizer friction linings . . . . . . . . . . . . . . . . . . . 24
2.3.8 Synchronizer manufacturing and tolerances . . . . . . . . . . 26
2.3.9 Synchronizer failure modes . . . . . . . . . . . . . . . . . . . 27
2.3.10 Synchronizer testing . . . . . . . . . . . . . . . . . . . . . . . 29
2.4 Maneuvering system . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.5 Other means to synchronize a gearbox and novel uses of synchronizers 31
2.6 Lubrication regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.7 Sliding contacts and surface temperature . . . . . . . . . . . . . . . . 33
2.8 Contact modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3 Research methodology 39
3.1 Fluid-structure interaction model to simulate pre-synchronization . . 39
3.2 Thermomechanical model to simulate main synchronization . . . . . 41
3.2.1 Contact properties used in this thesis . . . . . . . . . . . . . 43
3.2.2 Verification and validation . . . . . . . . . . . . . . . . . . . . 44
3.2.3 1D thermal model . . . . . . . . . . . . . . . . . . . . . . . . 46
vi
CONTENTS vii
3.2.4 Integrating simulations and physical tests . . . . . . . . . . . 48

3.3 Friction model for main synchronization . . . . . . . . . . . . . . . . 48
3.4 Speed step test to determine limiting load . . . . . . . . . . . . . . . 49
3.5 CFRP material data and simulations . . . . . . . . . . . . . . . . . . 50
3.6 Tolerance effects and cone angle optimization . . . . . . . . . . . . . 56
4 Results 59
4.1 Pre-synchronization simulation . . . . . . . . . . . . . . . . . . . . . 59
4.2 Main synchronization simulation . . . . . . . . . . . . . . . . . . . . 62
4.3 Friction model for main synchronization . . . . . . . . . . . . . . . . 63
4.4 Molybdenum coated synchronizer load limit . . . . . . . . . . . . . . 66
4.5 Results of CFRP simulation and testing . . . . . . . . . . . . . . . . 68
4.6 Effect of manufacturing tolerances and optimal cone angle . . . . . . 69
5 Summary of appended papers 75
6 Conclusions, discussion and future work 79

6.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.2 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.3 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
References 85
List of appended papers
This thesis consists of a summary and the following seven papers:
Paper A
Daniel Häggström, Ulf Sellgren, Stefan Björklund. “Robust pre-synchronization in
heavy truck transmissions”. Presented at the International Gear Conference, Au-
gust 26-28 2014, Lyon, France.
The author performed the simulations and the data evaluation. The author did
most of the paper writing.
Paper B
Daniel Häggström, Wiktor Stenström, Ulf Sellgren, Stefan Björklund. “Parameter
study of the thermomechanical performance of heavy duty synchronizers”. Pre-
sented at the 15th International VDI Congress – Drivetrain for Vehicles 2015, June
16-17 2015, Friedrichshafen, Germany, VDI-Berichte 2256, p. 597-620.
The author performed the simulations and most of the data evaluation. The author
did most of the paper writing.
Paper C
Daniel Häggström, Wiktor Stenström, Ulf Sellgren, Stefan Björklund. “A verified
and validated model for simulation-driven design of heavy duty truck synchroniz-
ers”. Presented at ASME 2015 Power Transmission and Gearing Conference, August
2-5 2015, Boston, USA.
The author performed the simulations and the data evaluation. The author planned
and contributed to the experimental work. The author did most of the paper writ-
ing.
Paper D
Daniel Häggström, Pär Nyman, Ulf Sellgren, Stefan Björklund. “Predicting friction
in synchronizer systems”. Tribology International, 2016, 97: 89-96.
viii
CONTENTS ix
The author performed the simulations and contributed to the data evaluation. The
author planned and contributed to the experimental work. The author did most of
the paper writing.
Paper E
Daniel Häggström, Ulf Sellgren, Stefan Björklund. "Evaluation of synchronizer
loading parameters and their ability to predict failure". Proceedings of the Insti-
tution of Mechanical Engineers, Part J: Journal of Engineering Tribology, 2017,
232(9): 1093-1104.
The author performed the simulations and the data evaluation. The author planned
and contributed to the experimental work. The author did most of the paper writ-
ing.
Paper F
Daniel Häggström, Ulf Sellgren, Stefan Björklund. “Thermomechanical perfor-
mance of CFRP synchronizer friction liners”. Article submitted to scientific journal.
The author planned and contributed to the physical tests. The author performed
the simulations and the data evaluation. The author did most of the paper writing.
Paper G
Daniel Häggström, Ulf Sellgren, Stefan Björklund. “Optimization of synchronizer
cone angle with regards to manufacturing tolerances of cone roundness and cone
angle”. Article submitted to scientific journal.
Publications not included in this thesis

Daniel Häggström, Ulf Sellgren, Stefan Björklund. “The effect of manufacturing
tolerances on the thermomechanical load of gearbox synchronizers”. Presented at
CIRP Conference on Manufacturing Systems, May 16-18 2018, Stockholm, Sweden.
Procedia CIRP, 2018, 72:1202-1207.
Abbreviations and alternative part
naming
AMT Automated manual transmission

or Automatic mechanically engaged transmission
AT Automatic transmission
BL Boundary lubrication
CAD Computer-aided design
CEC Coordinating European Council
CFD Computational fluid dynamics
CFRP Carbon fiber reinforced polymer
CH Crawler high / Crawler gear high split
CL Crawler low / Crawler gear low split
CMM Coordinate measuring machine
CTE Coefficient of thermal expansion
DCT Dual clutch transmission
DMA Dynamic mechanical analysis
DSC Differential scanning calorimetry
EP Extreme pressure (lubricant additive)
FE Finite element
FEA Finite element analysis
FEM Finite element method
FL Full film lubrication
FSI Fluid-structure interaction
GMS Gearbox management system
GVW Gross vehicle weight
ISO International Organization for Standardization
HPF Heat partitioning factor
LCP Linear complementarity problem
LFA Laser flash analysis
ML Mixed lubrication
MT Manual transmission
MTF Manual transmission fluid
x
CONTENTS xi
PAO Poly-alpha-olefin
PDE Partial differential equation
RH Reverse high / Reverse gear high split
RL Reverse low / Reverse gear low split
RPM Revolutions per minute
SEM Scanning Electron Microscopy
TEI Thermoelastic instability
Alternative part naming

Latch cone is also known as blocker ring, baulk ring and synchronizer ring.
Inner cone is also known as synchronizer cone.
Driver is also known as hub.
Shift sleeve is also known as sleeve or collar.
The cone half angle is often called the synchronizer cone angle.
For synchronizers, a CFRP lining is often referred to as “woven carbon lining” or
simply “carbon lining”.
A molybdenum coated synchronizer is often referred to as a “molybdenum synchro-
nizer”.
A CFRP lined synchronizer is often referred to as a “carbon synchronizer”.
Notation
A Out-of-roundness amplitude [m]

AC Nominal cone area [m2 ]
E Young’s modulus [Pa]
E∗ Effective Young’s modulus [Pa]
ES Specific synchronization energy [J/m2 ]
F0 Applied load, simplified 3D FSI model
FAx Axial force applied on shift sleeve [N]
Fmax Maximum axial force from pneumatic actuator [N]
FSpring Spring system force [N]
ḞAx Axial force ramp [N/s]
GD Groove depth [m]
GH Groove height [m]
GW Groove width [m]
I Moment of inertia to synchronize [kg m2 ]
Iz Moment of inertia, component z [kg m2 ]
IR Reflected moment of inertia [kg m2 ]
K Thermal diffusivity [m2 /s]
L Contact length [m]
L1 Length of simulation domain, simplified 3D FSI model [m]
P Applied normal load [N]
Pmean Mean synchronization power (including cooling time) [W/m2 ]
PS Specific synchronization power [W/m2 ]
Pe Peclet number [-]
RB Radius of blocker teeth / shift sleeve contact [m]
RC Cone radius [m]
Ri Radial coord. of node i [m]
Ri (θ)N om Nominal coord. of node i [m]
T Temperature [K]
TC Cone torque [N m]
TI Index torque [N m]
TS Average surface temperature [K]
Ta Temperature, evaluation point a [K]
Tinitial Temperature at start of synchronization [K]
xii
CONTENTS xiii
U Sliding speed, simplified 3D FSI model [m/s]

W Width of simulation domain, simplified 3D FSI model [m]
c Initial contact distance [m]
cp Specific heat capacity [J/(kg K)]
h Node to node distance in 1D model [m]
hc Contact overclosure [m]
iz Gear ratio between component z and gear to be synchronized [-]
m Out-of-roundness order [-]
n Number of cones [-]
p Contact pressure [Pa]
pa Contact pressure, evaluation point a [Pa]
p0 Contact pressure at zero gap [Pa]
q̇ Heat flux density [W/m2 ]
t Time [s]
tCritical Critical time step [s]
ts Time step in 1D model [s]
tC Cycle time [s]
v Sliding speed [m/s]
α Cone half angle [rad]
αLatch Latch cone half angle [rad]
αInner Inner cone half angle [rad]
θ Anglular coordinate [°]
θB Blocker teeth angle [rad]
γ Heat partitioning factor [-]
η Dynamic viscosity [Pa s]
λ Thermal conductivity [W/(m K)]
λ∗ Thermal conductivity over contact [W/(m K)]
µ Coefficient of friction [-]
µB Coefficient of friction in blocker teeth /
shift sleeve contact interface [-]
µC Coefficient of friction in cone interface [-]
µCmin Lowest coefficient of friction (in cone interface)
that gives blocking function [-]
ν Poisson’s ratio [-]
ρ Density [kg/m3 ]
φ Relative starting position [°]
ω Relative rotational speed [rad/s]
ωC Rotational speed, coupling disc [rad/s]
ωS Rotational speed, shift sleeve [rad/s]
ωI Initial relative rotational speed [rad/s]
ωL Lowest considered rotational speed difference [rad/s]
(1 ,2 Material 1, 2)
Chapter 1
Introduction
Around 35 % of the total cost of ownership for a long haulage truck is fuel expenses
[1], and together with the increased focus on CO2 -reduction from authorities [2],
customers and the public [3], it is clear that energy efficiency is one of the main
drivers in truck development. In long haulage operation, roughly 20 % of the energy
losses can be attributed to the powertrain [1], and 80 % of the powertrain losses
arise in the engine [1]. The engine working conditions are heavily influenced by the
current gear ratio, which is a combination of available gear ratios in the gearbox,
rear axle ratio, wheel radius and gear selection strategy [4]. One way to increase
overall vehicle efficiency is to allow the engine to run at a lower speed with a higher
torque [5, 6]. To cope with the power and efficiency demands, such a powertrain
would require more frequent gear shifts [7, 8].
The higher torque from the engine requires gear wheels and a clutch that can
handle the increased torque, leading to a higher moment of inertia of the gearbox.
To combine good startability and low engine cruising speed, the total ratio spread
of the gearbox has to increase, leading to higher rotational speed differences at gear
changes unless more gears are added to the gearbox. As the gearbox is adapted to
future powertrain concepts, future synchronizers and gear shifting systems need to
handle more frequent gear shifts with both higher moment of inertia and rotational
speed difference to synchronize. With more frequent gear shifts, the performance,
i.e. shift times, becomes more important to the overall transportation time as well
as driving comfort. Additionally, increased number of gear shifts put synchronizer
and gear shift system service life and reliability in focus. Engine downspeeding can
partly offset the increased rotational speed to be synchronized. However, in critical
driving situation there is a risk of not being able to complete the gear shift before
the vehicle retardation is so severe that the new gear cannot be engaged without
the risk of engine stalling. In such situations, the gear shift will be performed at
high engine speeds and thus requires a synchronizer that can handle high speeds to
be synchronized.
1
2 CHAPTER 1. INTRODUCTION
The market trend from manual transmissions to 2-pedal∗ systems such as au-
tomated manual transmissions [9] and dual clutch transmissions [10] means the
synchronizer is controlled by actuators and software rather than being under direct
control of the driver. This means that the synchronizer load can be controlled and
an automated trade-off between service life and performance can be made. This
gives a good potential to increase the reliability of the synchronizers by avoiding
overload and misuse situations from e.g. an unintentional selection of the wrong
gear by the driver.
The average profit margin in EU road freight transport 2000-2006 was 3.13 %
[11]. A simplified analysis is to assume a yearly distance of 200 000 km, an average
fuel consumption of 25 l/100km, a fuel price of 1.5 EUR and that 35 % of the total
cost is fuel, suggests that running a truck costs about 215 000 EUR/year. 3 % of
215 000 EUR is less than the cost of replacing a synchronizer. This means that one
synchronizer failure risks eating up a whole years profit.
It is clear that the gear shifts affect the fuel consumption of and emissions
from a vehicle [12], and that there is a lot of potential energy savings that can be
achieved by optimizing engine, gearbox and control software together. It is also
clear that the synchronizer must not fail during the life of the truck. The working
conditions of synchronizers can also be controlled to a greater degree than what has
been historically possible. This gives the gearbox designers new challenges and new
opportunities. Currently, there exist a knowledge gap on how to relate thermome-
chanical conditions in the contact interfaces to synchronizer service life and gear
shift performance. Parameters such as specific energy to synchronize and specific
synchronization power are often used to describe synchronizer load. Such parame-
ters have limitations, which is shown in this thesis. To meet the new challenges and
make use of the new opportunities, detailed knowledge on thermomechanical condi-
tions and its affect on the synchronizer performance and reliability is needed. This
knowledge needs to be implemented in the development tools and methodologies.
1.1 Research objective and motivation

The main motive of this research is a strategy aiming at a more model-based and
simulation-driven process for synchronizer development and validation, i.e. to sig-
nificantly increase the amount of computer simulation in synchronizer development
and validation. The effect of increased use of simulation is potentially shorter time
to market and better product quality. To realize the goal of simulation-driven de-
sign, usable and reliable methodologies and models are needed. Some work is done
on synchronizer simulation, e.g. with SYNTEM [13] or finite element analysis [14].
Gear shift loads follows a distribution, where the primary differentiator between dif-
ferent gear shifts is the rotational speed to synchronize. Much of the work published
focuses on long-term wear of synchronizer from intermediate to high repeated loads
∗ Throttle and brake pedal, no clutch pedal

1.2. RESEARCH QUESTIONS 3
[14, 15], or on extreme loads which results in almost sudden failures [16], which is
a common cause of complete gearbox failures.
1.2 Research questions

The main research question for this thesis is:
How can the state change (i.e. behavior) of the contact interface in a synchro-
nizer be predicted during pre-synchronization and main synchronization?
The main research question is divided into the following sub-questions:
• RQ1: How can the frictional behavior at transition between pre-synchronization

and main synchronization be predicted?
• RQ2: How does the coefficient of friction depend on the operating conditions
during main synchronization, and how can it be modeled to allow for more
detailed systems simulations?
• RQ3: How do external loads as well as different characteristic design pa-

rameters and their interactions affect the nominal and local contact interface
temperatures during synchronization?
• RQ4: How does the service life of a molybdenum coated synchronizer depend
on the applied load, and how can the engineers developing synchronizers en-
sure that the synchronizer has sufficient service life?
• RQ5: How can the material properties of a CFRP friction lining be estimated,
and how do they influence the surface temperature?
• RQ6: Is there a surface temperature in a synchronizer with CFRP lining,

above which surface damages can be observed and accelerated wear and per-
formance degradation will occur?
• RQ7: How do manufacturing tolerances affect the synchronizers maximum

surface temperature during synchronization? Can the effects be reduced by
changing the nominal design?
• RQ8: How can thermomechanical synchronization models be validated?
The research has been divided into smaller parts that contribute to the overall
goal. In paper A, the oil film thickness during the pre-synchronization is assessed.
This gives an indication of whether the latch cone is able to block the gear en-
gagement at transition between pre-synchronization and the main synchronization
phase. The problem description is thoroughly described in section 2.3.3 - Prevent-
ing unsynchronized gear engagement and 2.3.4 - Pre-synchronization phase, as well
as section 2.6 - Lubrication regimes for general information. In paper B and paper
C, a thermomechanical simulation model is developed and validated. Validation
is performed using both surface and bulk temperature measurements, as well as
visual inspection of wear marks compared with the highest contact temperature
areas indicated by the simulation. The thermomechanical loading of a synchronizer
is described in section 2.3.5 - Main synchronization phase, and background infor-
mation of sliding contact and thermomechanical loads are available in section 2.7 -
Sliding contacts and surface temperature. In paper D, a friction model is developed
for molybdenum coated synchronizers based on a combination of physical testing
and simulations. The model can be used to evaluate e.g. rigid body motion of
synchronizer components [17]. In paper E, the load limit for molybdenum coated
synchronizers is investigated, i.e. the load where the synchronizer fails after a few
cycles. It is shown that the service life of the synchronizer as a function of load, in
this case rotational speed to synchronize, is highly non-linear, where a slight reduc-
tion in load yields a significantly longer service life. In paper F, the temperature
dependent material properties of a carbon fiber reinforced polymer (CFRP) lining
is estimated. A material data sensitivity analysis is performed to evaluate the rel-
ative importance of each material parameter. Simulations using the material data
suggest at what temperature levels hot spots, accelerated wear and a reduction in
coefficient of friction start to appear. In paper G, the effect of manufacturing toler-
ances on the local thermomechanical load is assessed for both molybdenum coated
and CFRP lined synchronizers. Based on data for real manufacturing geometric
deviations, distributions of thermomechanical loads are estimated. It is shown that
it is possible to reduce the maximum surface temperature by introducing a rela-
tive angle between the cones contact surfaces. The optimum relative cone angle
is estimated based on measured synchronizer geometric deviations. This directly
influences the synchronizer population robustness. The contribution of the seven
papers to the overall goal is presented in Figure 1.1.
1.3 Thesis outline
Chapter 1 gives a short introduction to how gear shifts and synchronizers affect
the performance and reliability of the complete vehicle, and why more research is
needed on the topic. Chapter 2 gives detailed background information of synchro-
nization and gear shifts in general exemplified by a heavy duty truck gearbox and
its synchronizers. It also provides the theoretical background for topics needed for
this analysis, such as sliding contacts and contact modeling. Chapter 3 presents
the research methodology used in the thesis. Chapter 4 presents the results. In
chapter 5, the appended papers are summarized. Chapter 6 contains a discussion
about the work and results, ideas for future work, and the conclusions from this
thesis.
1.4. SWEDISH TRANSMISSION CLUSTER 5
Virtual development of
gear shifting system Research at Chalmers
University of Technology
in the same project
B
r
pe
Pa
C
A
er
p
er
Pa
F
p
er
Pa
er
E
p
p
Pa
Pa
er
G
p
Pa
er
p
Pa
Virtual development of Virtual assessment of

synchronizer functionality synchronization load and
and performance service life
Figure 1.1: Contributions to the overall goal by the appended papers.
1.4 Swedish transmission cluster

Sweden is one of the leading nations in development and manufacturing of trans-
missions for heavy duty vehicles. Swedish companies supplies gearboxes to four†
of the seven‡ leading European manufacturers of heavy duty commercial vehicles
(Gross vehicle weight, GVW >16 000 kg) with a combined European market share
(2014) of about 56 % [18]. In Sweden, more than 5000 persons work with develop-
ment or manufacturing of heavy duty transmission systems. To help maintain this
position, the Swedish transmission cluster was founded in 2011 as a collaboration
between Scania CV AB, Volvo Group Trucks Technology, KTH Royal Institute of
Technology and Chalmers University of Technology. The purpose is to support and
synchronize technology-neutral transmission related research as well as influence
education at both Master and Doctoral level. The cluster is partly financed by
† ScaniaCV AB supplies MAN Truck and Bus AG, Volvo Trucks supplies Renault Trucks
‡ The previously mentioned manufacturers as well as Mercedes-Benz Trucks, DAF Trucks NV
and Iveco.
the Swedish Governmental Agency for Innovation Systems, VINNOVA. This thesis
joins previous theses in the cluster such as investigation of churning losses and ef-
ficiency in gearboxes [19], lubricant flow and CFD methodology [20], running-in of
gear wheels [21] and surface stress evolution during running-in [22]. In addition to
this thesis, the current project includes development of models and a methodology
for Pareto optimization of synchronized gear shifts on a system level, i.e. the syn-
chronizer and its surroundings [17]. A project researching transmission vibrations
has also started [23, 24]. The cluster focuses on transmissions for heavy duty trucks
with European classification N3 (GVW >12 000 kg) [25], which is roughly equiva-
lent to US class 8 (GVW >33 000 lb, or about 15 000 kg) [26]. However, the results
might be transferable to other types of high-performing transmissions as well.
Chapter 2
Frame of Reference
This chapter provides background information to the presented research. Section

2.1 describes the general trend towards virtual product development, and shows
some advantages of computer simulations. In section 2.2, general gear shifting
principles, exemplified by a heavy truck transmission, are introduced. Section 2.3
provides an extensive background of the inner workings of a synchronizer, including
how to prevent gear engagement during asynchronous rotational speed and synchro-
nizer failure modes. Section 2.4 briefly presents an example of a pneumatic ma-
neuvering system, which applies the axial force to the synchronizer. In section 2.5,
other means to synchronize a gearbox and novel uses of synchronizers are presented.
Section 2.6 briefly presents the traditional lubrication regimes, and how they relate
to a synchronizer. In section 2.7, sliding contacts and thermomechanical instability
are presented, with an example from a molybdenum coated synchronizer. Finally,
in section 2.8, methods and challenges related to contact simulation are presented.
2.1 Simulation-driven product development

The customer demands on the product and the product complexity increases, and
stricter legislation such as emission and efficiency standards are introduced. Conse-
quently, the challenges, costs and risks in product development increase. To allow
for better products with shorter time to market, new development methodologies
and processes are needed. The trend in product development is to use computer
simulations to a large extent [27]. Systematic use of simulations during product
development is often referred to as model-based and simulation-driven design [28].
Concepts and solutions can be virtually assessed before prototypes are ordered.
The advantages compared to the old design-build-test method are many and the
potential effects on product quality and development time are huge. Product per-
formance can be improved by better knowledge of the product limitations and
operations, weight can be reduced by optimizing topology [29], reliability and ro-
bustness can be increased by identifying weak spots that can be improved [30],
7
8 CHAPTER 2. FRAME OF REFERENCE
acoustic emissions can be reduced by altering the stiffness of the product [31], and
so on. More iterations in the product development can be performed since sim-
ulations can be very time efficient, once you have the right models for the task,
compared to manufacturing and testing physical prototypes [27]. Cost can also
be reduced, since less physical prototypes and test rigs are needed. Simulations
can also give answers to questions on a phenomenon level that testing cannot due
to limitations in measurement techniques and feasibility of measuring all relevant
positions and all physical phenomena [32]. In short, better knowledge about the
product and its interactions with the surroundings can give a better product due
to better tools that assist development [33].
Advanced simulations can include several physical phenomena, i.e. multiphysics
effects, complex geometries and materials as well as local interactions between parts.
However, decisions based on incorrect simulations can have huge consequences on
performance and reliability of the finished product [34]. Therefore, model verifica-
tion and validation are important [35, 36].
Many physical phenomena require treatment of several independent variables
such as shape (between 1 and 3 dimensions) and time (for transient analysis), as
well as dependent variables such as temperature and displacement. The problem
can often be described by partial differential equations (PDE). While analytical so-
lutions to some PDEs exist, it is generally impossible to find an analytical solution
for real world engineering problems. However, the solutions can be approximated
by numerical methods [37]. The most commonly used is the Finite Element Method
(FEM, sometimes referred to as Finite Element Analysis, FEA, or Finite Element,
FE). The geometry, or simulation domain, is discretized into smaller elements. The
geometry can preferably be imported from the Computer-aided design (CAD) sys-
tems the design engineers use. The discretized geometry is often called “mesh”. An
example of original and discretized geometry is available in Figure 2.1. An approx-
imation of the solution is made in each discrete element, and then the elements
are patched together to a global solution [38]. For transient cases, the time is also
discretized. Numerical methods are used to find the solution [39, 40]. Since the
accuracy of the global solution depends on the per-element solution, the elements
have to be sufficiently small so that the local approximation is valid [41]. However,
smaller elements increase the number of elements, thus increasing computational
time.
2.2. GEAR SHIFT IN HEAVY TRUCK TRANSMISSIONS 9
(a) (b)
Figure 2.1: (a) Continuous CAD geometry and (b) discretized approximation of geometry
that can be used in FE simulations.
2.2 Gear shift in heavy truck transmissions

The following description is based on a Scania range-splitter gearbox and synchro-
nizer, but general enough to apply to most heavy duty gearboxes. Figure 2.2 shows
an example of a heavy duty truck gearbox. It consists of an input shaft (1), lay shaft
(2), main shaft (3) and output shaft (4) mounted in the gearbox housings (5-7).
The gearbox can be described as three serially connected gearboxes: the splitter,
the main gearbox and the range unit. The splitter consists of gear wheel pairs 8
and 9. The main gearbox consists of gear wheel pairs 11 for 1st gear, 10 for 2nd
gear, 9 for 3rd gear, 12 for the crawler gear, 13a for the reverse gear. The reverse
idler gear wheel is schematically shown as 13b. The range unit is a planetary gear,
where the ring gear is locked to the sun gear via 19a for high range gear, or to
gearbox housings 6 and 7 via 19b for low range gear. This gives 2 · 3 · 2 + 2 = 14
forward gears and 2 reverse gears, since crawler and reverse is not allowed on high
range gear. Table 2.1 shows the possible combinations.
5
6 7
15 12 19a 14 19b
16 10 11 17 13a
8 9 18
1
4
3
2
13b
Figure 2.2: Section cut of a heavy duty truck gearbox (Scania GRS905), including power
flow for gear 7 and close-up of splitter synchronizer. The reverse idler gear is schematically
shown.
Table 2.1: Gear number based on splitter, main gearbox and range position (Scania
GRS905).
Main gearbox Low range High range

Low split High split Low split High split
1 1 2 7 8
2 3 4 9 10
3 5 6 11 12
Crawler CL CH Not used Not used
Reverse RL RH Not used Not used
The clutch disc (not shown in Figure 2.2), which is mounted on the input shaft
(1), transfers the torque from the engine to the gearbox. The torque is transferred
via the active split gear wheel pair (8 or 9) to the lay shaft, and then to the main
shaft via the selected main gearbox gear (9-13) and to the output shaft via the
range gear (14, via 19a or 19b). The output shaft is connected to the propeller
shaft, which is connected to the wheels via the rear axle, half shafts and potentially
the hub reductions. The gear wheels on the lay shaft (2) are rotationally locked
to the shaft or integrated into the shaft itself. The gear wheels on the main shaft
(3) and input shaft (1) are free to rotate around their shaft. The gear wheels in
each gear wheel pair (8-13) are in constant mesh with each other. To transfer
the driving torque, the rotatable gear wheel is connected to its shaft by axially
2.3. SYNCHRONIZER BACKGROUND INFORMATION 11
displaceable shift sleeves (15-18). Note that gear wheel 9 is used for both splitter
and main gearbox gear, and if gear wheel 9 is engaged by both shift sleeve 15 and
shift sleeve 16 the direct-drive is engaged, and the torque is not transferred to the
lay shaft thus eliminating load dependent mesh losses and increasing efficiency. For
the unsynchronized reverse gear and crawler gear, a dog clutch (18) is used instead
of the synchronizer. The blue line schematically shows the power flow for gear 7.
The gearbox is available as both a manual transmission (MT) variant and as an
automated manual transmission (AMT) variant. For the AMT variant, the gear
lever and gear linkage is replaced by pneumatic actuators. For the driver, the truck
and gearbox is then operated similarly as a conventional automatic transmission.
As an example, assuming the retardation of the vehicle during gear shift is
negligible, a gear shift from 2nd to 3rd main gearbox gear, without any split or
range shift, would be:
1. Initial states. The wheels rotate at a constant rotational speed, which makes
the output shaft rotate at a constant rotational speed. In the example, the
range is not shifting, so the main shaft also rotates at a constant speed. The
rotational speed difference between the main shaft and the gear wheel to be
engaged is determined by the difference in gear ratio between current and
target gear as well as rotational speed of the main shaft.
2. Open the clutch. The transfered torque over the clutch is reduced so that
the elastic deformation of the propeller shaft and half shaft is reduced to avoid
oscillations in the gearbox. The gearbox is disconnected from the engine by
separating the clutch disc from the flywheel and pressure plate.
3. Engage neutral gear. Main gearbox shift sleeve is moved to neutral posi-
tion.
4. Synchronize the rotational speeds. Since a new gear ratio will be en-
gaged, there exist a relative rotational speed between the main shaft and the
gear wheel to be engaged. Thus the gear wheel has to be braked to the target
rotational speed, which means braking the lay shaft, all meshing gear wheels,
the input shaft and the clutch disc. Note that a downshift would require
speed increase of the previously mentioned components. Synchronizers can
be used to synchronize the relative rotational speeds. Some alternatives to
synchronizers are presented in section 2.5 - Other means to synchronize a
gearbox and novel uses of synchronizers.
5. Engage gear. Main gearbox shift sleeve is moved to the engaged position
for the target gear.
2.3 Synchronizer background information

Synchronizers have been used for almost 100 years [42], and represent one of the
biggest improvements made in the history of automotive transmissions [43]. It is a
mechanical machine element that synchronizes the rotational speed of the gearbox
during gear shifts and prevents gear engagement during asynchronous rotational
speeds. Figure 2.3 shows the single cone synchronizer used in the gearbox in Figure
2.2. The driver component is connected to the input shaft (1 in Figure 2.2) for
the splitter synchronizer (15) and to the main shaft (3) for 2nd /3rd and 1st gear
synchronizers (16, 17). The shift sleeve is mounted on the driver and can move
axially. The protruding blockers of the latch cone are mounted in the slots in
the driver. The coupling disc is mounted on the gear wheel, and the inner cone
is mounted in the holes of the coupling disc (note that this differs slightly for
double and triple cone synchronizers). A friction lining is applied to the latch
cone, in this case flame sprayed molybdenum. Sintered bronze, CFRP or carbon
based compounds are also commonly used [44, 45]. To transfer driving torque, the
coupling disc is locked to the driver by the shift sleeve. The teeth on the coupling
disc and the teeth in the shift sleeve have a “locking function”, which prevents gear
disengagement during torque transfer. The locking function is often in the form of
a backwards taper angle. The wire spring temporarily prevents gear engagement
at start of synchronization.
A gearbox generally contains single cone, double cone, and triple cone synchro-
nizers. The different synchronizers in the gearbox have different loads since they act
on different gears. At lower gear, the inertia to synchronize is significantly higher
than at higher gears [46]. For example, the gearbox in Figure 2.2 has an inertia
of 2.62 kg m2 at 1st /7th gear, and 0.44 kg m2 at 5th /11th gear. However, at lower
gears the speed differences are generally lower, for example a shift from 9th to 7th
gear with an engine speed of 1500 RPM after the shifts yields a speed difference
of 181 RPM (formulas available in reference [47]), while a shift from 9th to 11th ,
with an engine speed of 1500 RPM before the gear shift, gives a speed difference
of 432 RPM. Something that complicates things further is that the driver behavior
or the Gearbox Management System (GMS) gearshift strategy likely is different at
low and high gears. However, it is clear that a lower gear has a higher inertia to
synchronize, while a higher gear generally has a larger initial speed difference to
synchronize. Preferably, the synchronization time for each gear should be similar
and the synchronizers are used to their limit. This means that different synchroniz-
ers are needed for the different gears. Figure 2.4 shows the Scania modular system
for synchronizers where a single cone, double cone and triple cone synchronizer can
be built from common parts and parts with small differences between the variants.
The shift sleeve, driver and wire spring are common parts between the systems.
The coupling disc is common between the single and double cone synchronizer, and
similar but with an integrated cone for the triple cone synchronizer. The latch
cones are similar, but the lugs connecting to the connected cone marked in green
as well as the blocker teeth angle differ. The green cone has no cone interface on
the inner diameter for the double cone synchronizer. To the reduce the complexity
in both simulation and testing, this thesis focuses on single cone synchronizers.
Figure 2.3: Example of synchronizer, Scania’s current (2018) single cone synchronizer.
(a) (b) (c)
Figure 2.4: Scania modular synchronizer system. (a) Single cone synchronizer. (b) Double
cone synchronizer. (c) Triple cone synchronizer.
2.3.1 Related work

Synchronizers
Synchronizer research can take many forms. Lovas et al. divided the synchroni-
zation process into 8 phases and simulated the entire synchronization process [48].
Disturbances in the form of stick-slip between the cones as well as in the sleeve
splines were introduced and analyzed. The double bump phenomenon was inves-
tigated. Paffoni et al. investigated the oil squeeze during the pre-synchronization
phase [49], and concluded that radial grooves appear to be better than circumfer-
ential grooves to evacuate the oil between the cones. Moir investigated the gear
shift quality, namely the shift force, shift feel, and shift definition [50]. This work
was later used to develop the Ricardo Gearshift Quality Assessment (GSQA) tool.
Kinugasa et al. measured and simulated the contact surface temperature of a
brass synchronizer [16]. They found that at over 180 ◦C, the wear rate as well as the
risk for clashing increased (i.e. gear engagement at unsynchronous speed, see 2.3.9
- Synchronizer failure modes for more information). Lösche et al. experimentally
investigated the wear of a molybdenum coated synchronizer for light duty truck as
well as a brass synchronizer for passenger cars [15]. They developed a semi-empirical
wear model using a “duty parameter” that depends on sliding speed, contact pres-
sure and rotational energy to synchronize. They found good correlation between
measured and calculated wear, and hypothesized that this “duty parameter”’ re-
flects the contact surface temperature. Spreckels found that for molybdenum coated
synchronizers for light duty trucks, both coefficient of friction and wear rate depend
on the loading [51]. He simulated the contact temperature on grooved synchronizers
and found that the temperature correlates well with the onset of heavy wear. Poll
and Spreckels also investigated the effect of cone angle differences on molybdenum
coated synchronizers [52], and found that small angle differences can have a large
impact on the contact surface temperature. Acuner et al. simulated the effect of
small cone angle differences for carbon lined synchronizers, and verified by physical
testing that a synchronizer with a small angle difference had significantly longer
service life than the serial design [53]. Neudörfer performed a similar study as
Spreckels, but expanded it to include both molybdenum coatings, sintered friction
linings, and carbon linings for passenger cars [14]. He found that the risk of clashing
increased as the carbon lining exceeded the thermal decomposition temperature.
Wanli et al. tested a molybdenum coated synchronizer until failure and found
that the surfaces roughness had decreased from Rq 0.45 µm and 0.72 µm to 0.24 µm
and 0.33 µm for the synchronizer ring and cone, respectively. Acuner tested different
kinds of carbon based linings under high and very high loads [13]. The FVA∗ -
developed program SYNTEM as well as FE models were used to simulate the
surface temperature.
∗ Forschungsvereinigung Antriebstechnik, the (German) Research Association for Drive Tech-
nology [54]
Several studies have also been performed to determine the influence of the lu-
bricant. Bouffet developed a pneumatically actuated synchronizer tribometer and
tested brass synchronizers [55]. Two design of experiments were performed, one at
low load and one at high load, and the coefficient of friction was evaluated at dif-
ferent sliding speeds. In both cases, the synchronizer type was the most important
parameter, followed by the oil viscosity and friction modifier. Brown et al. tested
three different friction modifier systems using the same PAO base oil and EP/anti-
wear package for brass, molybdenum coating, phenolic and two kinds of carbon
friction lining for the synchronizer [56]. Three parameters, namely Durability, Shift
quality and Wear, were compared. The Durability parameter was related to the
friction changes over 10 000 cycles. The Shift quality parameter was based on the
friction-speed dependency, where a “flat” profile is advantageous. The Wear pa-
rameter was determined by measuring the weight loss after the test. It was shown
that the frictional behavior when using different lubricants is strongly dependent
on lining material, especially when comparing metallic to non-metallic linings. The
wear was in most cases low and thus hard to differentiate between the test cases.
A follow-up study is performed with focus on analyzing the surfaces with different
surface analysis techniques [57]. Auger Electron Spectroscopy (AES) and argon
sputtering were used to determine elemental concentration at different depths for
brass and molybdenum coated synchronizers. It is shown that the lubricant that
generated the thickest tribofilm had best frictional and wear behavior for brass
synchronizers, but no such correlation was found on molybdenum coated synchro-
nizers. Since AES could not be used on non-conductive samples, the phenolic and
carbon synchronizers were analyzed using Scanning Electron Microscopy (SEM)
and Energy Dispersive Spectroscopy (EDS). The authors claimed that phenolic
and carbon synchronizer linings are becoming increasingly popular due to their low
interaction with the lubricant. The EDS elemental maps suggest that the silicon
supporting matrix had reacted with the lubricant, which might be the reason for
the different frictional behavior when using the different lubricants. Gangvekar and
Deshmane resolved early (<10 000 km) clashing by grit blasting brass synchroniz-
ers for a passenger car. A 20 % higher coefficient of friction was achieved, better
friction stability and higher wear resistance was achieved [58].
Walker wrote an extensive thesis on shift dynamics of dual clutch transmission
(DCT), analyzing gear shift on a systems level [59], including hydraulic system,
drag torque, and feedback control strategies. Alizadeh et al. designed a closed-
loop control system for a solenoid actuator to keep the synchronizer in the mixed
lubrication regime to minimize wear [60]. When using the controller in a laboratory
scale test rig, the synchronizer operated in the boundary lubrication regime for
220 ms and in the mixed lubrication regime for 1420 ms, while the uncontrolled
system operated for 830 ms and 470 ms in the boundary and mixed lubrication
regime, respectively.
M’Ewan has an interesting section on the effect of vehicle scale on synchronizer
design [61], and found that synchronizers for larger vehicles are more “difficult”
to design. He concludes that if a vehicles size is increased by a factor n, the
synchronization torque must increase by n3 to achieve similar synchronization time.

It is proposed that the maximum cone size increases approximately linearly with
vehicle size, and thus the axial force needs to increase by a factor n2 . While a
modern heavy duty truck is likely outside the domain where the assumptions are
valid, it shows that synchronizer design is vehicle class specific, and the bigger the
vehicle the bigger the demands are. Table 2.2 presents some differences between
single cone synchronizer dimensions and loading between an European heavy duty
truck and passenger cars. Data for passenger car synchronizers are taken from
the CEC L-66-99 [62] synchronizer test standard, which likely represents relatively
tough operations to reduce test time. These values should not be interpreted as
general values for all variants and applications, but as an approximate comparison.
Table 2.2: Comparison between truck and passenger car single cone synchronizer dimen-
sions and load.
Unit Truck Passenger car Ratio

Cone diameter mm 170 80 2.1
Cone width mm 12.5 8 1.6
Cone area mm2 6500 2000 3.3
Inertia kg m2 0.4 0.04 10
Speed difference RPM 500 1100 0.5
Sliding speed m/s 4.5 5.3 0.8
Force N 2500 720 3.5
Contact pressure MPa 4 3 1.3
Life expectancy Cycles 1 000 000 200 000 5
Gearbox type − 90 % AMT Mostly MT −
Wet clutches
Wet clutches are in many regards similar to synchronizers. Saito et al. tested
the wear behavior of paper based wet clutches, and simulated the temperature
distribution [63]. They found that the initial high, transient wear was governed by
the contact pressure and the work energy, while the lower steady state wear after
some thousand cycles was governed by the maximum temperature and work energy.
Ohkawa et al. investigated how the elasticity of the friction material affected the
dynamic coefficient of friction as well as the highest manageable load [64]. The load
was evaluated by the newly defined Energy-Power Absorbing Capacity, which is the
product of the (specific) kinetic energy during one engagement and the maximum
(specific) absorbed power during that engagement, with the unit J/cm2 ·kW/cm2 .
It was shown that for lower apparent elastic modulus, the dynamic coefficient of
friction is higher than for high modulus materials. Four different types of material
(elastomerics, paper, graphitics and sintered bronze friction lining) follow a linear
trend when plotted in a lin-log relationship (linear y-axis representing the coefficient
of friction, logarithmic x-axis representing the apparent elastic modulus). For the
load capacity, paper based linings seem relatively insensitive to changes in apparent
elastic modulus, while the other material types have a strong load capacity-apparent
elastic modulus relationship. It was found that the hot spot size was proportional
to the inverse of the apparent elastic modulus. It was believed that the reason is the
real contact area depends on the apparent elastic modulus. Okabe et al. tested wet
clutches for forklift trucks, and developed simplified thermal models by mapping
the measured temperature to the absorbed energy and the absorbed energy rate
[65]. Additionally, a T-N (temperature-cycles to damage) curve was developed for
the friction material, and a S-N (stress-cycles to damage) curve was developed for
the mating plate. Under 250 ◦C, the mating plate deformation occurred before
the friction material damage, and over 250 ◦C friction material damage occurred
before mating plate deformation. Usage data from end users was used to identify
damaging situations, and it was clear that certain usage patterns promoted either
mating plate deformation, friction material damage, or both. Hirano et al. tested
wet clutches using a speed step test with 9 cycles per step [66]. The formation of hot
spots was found by comparing the scatter in the separator plate temperature under
the 9 cycles, as well as the scatter in the torque curve for the 9 cycles. This was
verified by visually comparing plates before the step the indicated
Rt hot spots and
after the steps of indicated hot spots. Parameters Qs1 = 0 qmax q̇dt (cumulative
energy from initial engagement to the time of maximum power) and Vc , the critical
speed for formation of hot spots, were defined. The “Region of Hot spot formation”,
based on Qs1 and VC , was also defined. A new friction material was developed with
improved hot spot resistance.
Marklund et al. investigated friction characteristics in wet clutches using a
limited slip clutch (LSC) test rig [67]. The coefficient of friction from the LSC
was then expressed as a function of the sliding speed and four coefficients C1 to
C4 , where the coefficients depend on the temperature and applied axial load. The
model is valid in the boundary lubrication regime. Marklund and Larsson tested
wet clutch friction characteristics in a pin-on-disc test rig. The pin was modified
to act as a holder for a small sample of the friction disc [68]. A friction model
where the friction depends on the sliding speed and the interface temperature was
developed. The tests showed good correlation with full scale wet clutch tests. Saito
et al. developed a method to predict the total torque from a wet clutch installed in
an automatic transmission (AT), including effects from gear misalignments [69]. A
good correlation between measured and simulated torque was found, even for large
gear misalignments and clutches from high-torque AT.
Rank and Kearsey showed a general ranking of strengths and weaknesses of
molybdenum coating, paper based linings, sintered friction lining and carbon fric-
tion linings for limited slip differential applications [70]. Four different types of
carbon linings were also compared, namely woven, CVD, non-woven (composition)
and two-ply (where a conventional paper type friction material is layered with a
surface of almost pure carbon). Generally, the authors suggest that carbon linings
have good or very good properties compared to the other materials, except for cost
and friction level. Different carbon linings are tested both for frictional behavior
and durability. Six different oils were tested, and it was shown that the oil sensi-
tivity is relatively low. Oldfield and Watts tested the low speed (<2.5 m/s) friction
behavior for three different carbon fiber materials and compared them to a Kevlar
(i.e. aramid fiber) based and a paper based material [71]. Several different lubricant
formulations were used, namely an unformulated base stock as reference and the
base stock with 7 different “friction enhancers”’ and 7 different friction modifiers.
It was concluded that the carbon linings had better frictional properties when using
the base stock compared to the non-carbon linings, but worse frictional properties
when using a conventional automatic transmission fluid. The results show that
different lubricants are needed to maximize the benefits of the carbon linings.
2.3.2 Different phases of synchronization

Many definitions exist for synchronization phases [48, 72, 73, 74, 47]. Figure 2.5
shows in-truck measurements of an automated manual transmission gear shift, as
well as the teeth positions for five different synchronization phases used in this
thesis:
1. Neutral. Initially, the shift sleeve is in the neutral position (around 50 %

of total stroke length in the pneumatic actuator, since there is an in-gear
position on both sides of the neutral position).
2. Pre-synchronization. The shift sleeve leaves the neutral position to ener-

gize the spring system. The spring system temporarily prevents gear engage-
ment, i.e. the spring system force FSpring exceeds the axial force applied on
the shift sleeve FAx , as explained in section 2.3.4 - Pre-synchronization phase.
3. Main synchronization. The rotational speed is synchronized. During this

phase, the latch cone blocks the gear engagement. Note that the position
of pre-synchronization and main synchronization is approximately the same.
The shift sleeve does not move significantly in the axial direction during main
synchronization. A small movement can be seen in the measured data, but
it is due to initial clearances closing and elastic deformation of components.
The main synchronization phase is further described in section 2.3.5 - Main
synchronization phase.
4. Blocker release. Synchronous rotational speed is reached, and the latch

cone blocking function is released.
5. Gear engagement. The shift sleeve is moved to an engaged position. Since

the relative angular position of the shift sleeve and the coupling disc is stochas-
tic, the shift sleeve might or might not come into contact with the coupling
disc teeth [47, 75].
Synchronization phases
(a)
(b) (c) (d) (e) (f)
Figure 2.5: (a) Stroke position, main shaft rotational speed, lay shaft rotational speed
and calculated gear wheel rotational speed during synchronization, measured in truck.
Synchronization phases schematically shown. (b) Phase 1: Neutral. (c) Phase 2: Pre-
synchronization, FAx < FSpring . (d) Phase 3: Main synchronization, FAx ≥ FSpring . (e)
Phase 4: Blocker release. (f) Phase 5: Gear engagement.
Many synchronization process definitions also include gear disengagement before

neutral position phase, and a free-flight phase after leaving neutral position and
before entering pre-synchronization phase and a teeth-teeth contact phase during
gear engagement. In this thesis, the pre-synchronization and the main synchroni-
zation phases have been studied.
2.3.3 Preventing unsynchronized gear engagement

To allow synchronization of the rotational speed, gear engagement must be avoided
during synchronization. The latch cone is mounted on the driver component with
a small rotational clearance. When the latch cone is positioned in either of its
extreme positions as shown in Figure 2.6a, the teeth block the axial motion of the
shift sleeve, as shown in Figure 2.6b, preventing gear engagement. Two torques act
on the latch cone: the cone torque, TC , and the index torque, TI [76]. The cone
torque arises from the friction between the latch cone and the inner cone, and seeks
to reduce the rotational speed difference between the cones to zero as well as to
move the latch cone towards its extreme position. The cone torque is [76, 77]
n
X µCi · RCi
TC = FAx · (2.1)
i=1
sin(αi )
n is the number of cones, FAx is the axial force acting on the latch cone, µC is
the coefficient of friction between the cones, RC is the radius of the cones and α
is the cone half angle. The index torque, which seeks to rotate the latch cone into
its center position so that the shift sleeve can pass the latch cone teeth, can be
calculated by [76, 47]

1 − µB · tan(θB )
TI = FAx · RB · (2.2)
µB + tan(θB )
where RB is the radius of the teeth-teeth contact, θB is the angle of the blocker
teeth and µB is the coefficient of friction in the teeth-teeth contact. During synch-
ronization, the inequality TC > TI has to be fulfilled to ensure that the gear cannot
be engaged. The geometric definition of RC , RB , α and θB can be found in [47].
µB has a lower contribution to the torque balance than µC [78]. The lowest value
for the coefficient of friction, µCmin , for a single cone synchronizer that still blocks
the gear engagement, can be determined by

RB 1 − µB · tan(θB )
µCmin = · sin(α) · (2.3)
RC µB + tan(θB )
(a) Rotational clearance (b)
Shift sleeve
Latch cone,
blocking position
RB
Driver
Coupling disc
TC TI
Teeth contact, blocked engagement
Figure 2.6: (a) Rotational clearance between latch cone and driver. Note that TC and TI
act on the latch cone. (b) Latch cone in blocking position. Note the overlap of the teeth.
2.3.4 Pre-synchronization phase

At the beginning of synchronization, the conical surfaces are covered by lubricating
oil and there is a risk that the synchronizer is in the full film lubrication regime
(presented in section 2.6 - Lubrication regimes). This means that there is a risk
µC is lower than µCmin , meaning the latch cone cannot block gear engagement.
Therefore, another system to prevent gear engagement is needed. For the synch-
ronization system studied here, a wire spring is used as pre-synchronization spring
system. Another common design is to use radial spring-loaded plunges or detents
[79, 80]. To move the shift sleeve axially past the pre-synchronization position, the
spring has to be compressed, which requires a certain axial force. Before this axial
force is reached, gear engagement is prevented. The process is schematically shown
in Figure 2.7. The oil film thickness decreases as the oil is squeezed out from the
contact interface. The coefficient of friction, here schematically shown, increases as
the lubrication regime transitions from full film lubrication to boundary lubrica-
tion. At the start of pre-synchronization, the spring system needs to prevent gear
engagement. When the coefficient of friction exceeds µCmin , the latch cone can
block gear engagement.
Region where spring system Region where blocking is

needs to prevent gear engagement performed by latch cone
Schematic oil film thickness [m]
Schematic µC
Possible oil film
when leaving µCmin
neutral position
FL ML BL
Time [s]
Figure 2.7: The two different mechanisms to prevent gear engagement plotted against oil
film thickness and schematic coefficient of friction. The green lines schematically represent
transitions between lubrication regimes.
The transition between pre-synchronization phase and main synchronization

phase occurs when the axial force applied to the shift sleeve, FAx , exceeds the
spring system force, FSpring . Some rotational speed synchronization can occur
during pre-synchronization, since even the low value for the coefficient of friction
in the transition between the mixed lubrication regime and the full film lubrication
regime gives a small synchronization torque. However, when the transition to the
main synchronization phase is complete, the latch cone must be able to block gear
engagement since the spring system can no longer prevent gear engagement.
2.3.5 Main synchronization phase

During the main synchronization phase, the rotational speed difference is reduced
to zero. The lubrication regime is generally boundary lubrication. To be able to
engage the gear when synchronization is finished, the axial force has to be higher
than the spring system force. The coefficient of friction needs to be high enough
for the latch cone to block gear engagement. A high coefficient of friction also
leads to good synchronization performance, i.e. low synchronization time. The
acceleration or braking of a gearbox subsystem leads to frictional heating of the
contact interface. The following external parameters have been identified as the
main loading parameters during synchronization:
• Rotational speed difference to synchronize [14, 51].
• Axial force during synchronization [14, 51]. The transient axial force from a
pneumatic actuator can be approximated by
(
ḞAx · t t ≤ Fmax /ḞAx
FAx = (2.4)
Fmax t > Fmax /ḞAx
ḞAx is the axial force ramp, t is the time and Fmax is the maximum axial
force the actuator can deliver. For a manual transmission, the axial force is
based on the drivers input, which can be hard to describe mathematically
since every driver and gear shift is different [50].
• Moment of inertia to synchronize. The moment of inertia of every component

that will be synchronized is reflected to the gear wheel to be synchronized by
X
IR = Iz · i2z (2.5)
IR is the reflected moment of inertia, Iz is the moment of inertia of component

z, and iz is the gear ratio between component z and the gear wheel to be
synchronized [14, 47, 76]. Generally, the moment of inertia to synchronize, I,
is equal to IR .
• Acceleration of the shaft during gear shift due to vehicle acceleration. This
can either assist or counteract synchronization. An example is available in
Figure 2.8, where the transient rotational speed difference is affected by the
acceleration of the shaft.
• Drag torque, which either assists or counteracts synchronization depending

on up- or downshift, since the drag torque seeks to reduce the rotational speed
of the affected component [47].
• Clutch drag, i.e. that the clutch transfers torque from the flywheel when
the clutch is supposed to be disengaged. This can either assist or counteract
synchronization, depending on flywheel and clutch disc rotational speeds as
well as up- or downshift.
• Shift frequency and gearbox/oil temperature, which affect the initial temper-
ature of the synchronizer [14].
Effect of vehicle acceleration on synchronizer

1,500
Rotational speed [RPM]
1,000
500 Gear wheel

Shaft, constant speed
2
Shaft, acceleration -50 rad/s
0
0 0.04 0.08 0.12 0.16 0.2
Time [s]
Figure 2.8: Effect of main shaft acceleration on synchronization time and transient speed
difference. In this case, the main shaft acceleration due to vehicle acceleration increases
the synchronizer load.
These external loading parameters can be combined into what Lösche et al. called
“duty parameters” [15]. In this thesis, these following duty parameters are consid-
ered:
• Specific energy to synchronize, i.e. energy per (nominal) unit area.

• Specific synchronization power, i.e. power per (nominal) unit area.
• Surface temperature or surface temperature increase.
These duty parameters are defined in paper E.
2.3.6 Synchronizer wear and wear gap

The maximum permissible wear in a synchronizer, disregarding secondary effects
such as wear particles damaging bearings or clogging the oil filter, is determined
by:
• Changes in frictional behavior due to wear, which leads to loss of blocking

function and thus clashing. This ranges from changed surface texture to
complete removal of friction lining.
• The wear gap, shown in Figure 2.9. Wear, as well as plastic deformation,
causes the axial position of the latch cone during the main synchronization
phase to move towards the coupling disc. If the latch cone touches the cou-
pling disc, the contact pressure between the latch cone and the inner cone is
reduced, with the extreme case being a gap between the cones. This would
lead to a lower cone torque and clashing due to loss of blocking function [79].
To reduce installation space and reduce time between blocking release and
engagement, the wear gap shall be as small as possible. Note that the wear
gap also has to accommodate for geometrical deviations, i.e. tolerances. A
double or triple cone design needs a larger wear gap, since the total wear as
well as total acceptable geometric deviations are larger.
Figure 2.9: Synchronizer wear gap, i.e. distance between latch cone and coupling disc.
Note that the wear gap has to account for both wear, tolerance deviation and plastic de-
formation.
2.3.7 Synchronizer friction linings

Many different friction linings are used in modern synchronizers [81, 82, 83, 84]. For
lower loads, brass can be used. For low-intermediate loads, molybdenum is used,
and sintered friction linings can be used for intermediate-high loads. For high and
very high loads, carbon based linings are used. For special cases such as a reverse
gear synchronizer (which is intended to be used during stand-still), steel-on-steel
synchronizers [81] can be used. Carbon based linings include:
• Woven carbon fiber reinforced polymer (CFRP) [83].

• Two-ply material, where e.g. a highly conductive material is placed at the
surface of a conventional material [70, 85].
• Paper-based friction materials, where reinforcing fibers, binder, filler and fric-
tion modifier is mixed in a process similar to paper-making [86, 87].
Two friction linings are investigated in this thesis, namely molybdenum and a com-
mercially available woven CFRP with a phenolic resin and 50 % fiber content by
weight. Phenolic resin is obtained from the reaction of phenol and formaldehyde,
and is an amorphous thermoset polymer. The atomic structure of phenol, formalde-
hyde and phenolic resin is schematically shown in Figure 2.10. Phenolic resins are
commonly used as matrix material for frictionally heated systems due to its high
thermal stability [88, 89, 90]. Carbon fiber has the same basic structure as graphite,
which consists of layers of hexagonal patterns. In graphite, these are highly ar-
ranged and stacked in parallel, but in carbon fiber the layers are amorphous, folded
and interlock each other [91, 92]. The atomic structure of graphite/carbon fiber is
shown in Figure 2.11. A cross section of the CFRP lining is shown in Figure 2.12,
where the CFRP lining is molded in epoxy, polished, and the image is captured
using an optical microscope. Figure 2.12a shows the cross section with annotation.
The green lines shows the top and bottom of the lining, i.e. outside of the green
lines, we can see the epoxy holding the sample. The fibers appear black, while the
phenolic resin (as well as the epoxy) appears dark gray. The red and blue areas
show the two different tow directions, where the 2-direction is normal to the picture.
Figure 2.12b shows the original picture, without annotations, for reference.
OH OH OH
O
CH2 CH2 OH
C
H H
(a) Phenol (b) Formaldehyde (c) Phenolic resin
Figure 2.10: Schematic molecules for phenol (a), formaldehyde (b) and phenolic resin (c).
Phenolic resin is only partially shown, and can consist of large molecules [88].
Graphite/Carbon fiber
Figure 2.11: Schematic molecule structure of graphite and carbon fiber [91].
(
a)
Tow,
1-d
ire
cti
on Fi
ber
s
Ep
oxy Tow,
2-d
ire
cti
on Ma
tri
x
(
b) 500
µm
Or
igi
nal
Figure 2.12: Cross section of CFRP lining molded in Epoxy. (a) With annotation showing
tow directions. (b) Original.
2.3.8 Synchronizer manufacturing and tolerances

All physical parts have geometric imperfections, ranging from surface roughness to
macroscopic form deviations. The maximum allowed deviations are often referred
to as tolerances. International standards such as ISO 1101 [93] define how these
deviations shall be defined on drawings and evaluated on the physical product in
e.g. a coordinate measuring machine (CMM). Relevant tolerances for a synchro-
nizer include cone roundness, cone angle/cone angularity, run-out between teeth
and cone, cone straightness as well as general dimensions and angles of the part.
The surface texture of the cones is also a critical parameter. The choice of manufac-
turing process, manufacturing equipment and manufacturing parameters determine
the shape and magnitude of these geometrical deviations. Figure 2.13 shows the
manufacturing process for both the latch cone and the inner cone investigated in
this thesis.
Latch cone Inner cone
Forging of blank Use pipe as blank
Coining of functional surfaces Machine cone and lugs
Induction hardening of teeth Case hardening
Turning Hard turning of cone
Molybdenum CFRP
Shot peening Stamp CFRP strip
Bond CFRP to
Flame spraying
cone with adhesive
Figure 2.13: The manufacturing process of the latch cone and the inner cone.
2.3.9 Synchronizer failure modes

The main failure mode of a synchronizer is clashing, which is when the gear is
engaged with a remaining rotational speed difference, as shown in Figure 2.14. The
symptoms of clash are a “grinding” noise from the engagement teeth, and high
impulses on the engagement teeth, which will lead to plastic deformation and wear,
as shown in Figure 2.15. Clash can be caused by several issues, as shown in Figure
2.16. Full clash, i.e. when no synchronization of rotational speeds is performed
before the clash, can arise from an inadequate pre-synchronization phase that never
allows the latch cone to block the gear engagement. Wear of the contacting surfaces
that alters the frictional properties of the synchronizer can also cause clash, since
the latch cone cannot block gear engagement if the coefficient of friction is too low.
Sudden rotational speed changes in the powertrain, from e.g. clutch disengagement
during torque transfer or disturbances from the vehicle’s anti-lock braking system,
can temporarily reduce the indexing, i.e. overlap between latch cone blocker teeth
and the shift sleeve teeth, allowing for gear engagement. This can cause both full
or partial clash, i.e. when some synchronization of rotational speeds has occurred.
High drag torque from primarily the lay shaft, which is partly immersed the oil,
or torque transferred over the clutch can create a rotational speed difference after
synchronous rotational speed is reached but before the gear is engaged, which is a
form of clash called desynchronization [76].
ωS
ωC
Figure 2.14: Schematic description of clash. The rotational speed of the shift sleeve, ωC ,
is greater than the rotational speed of the coupling disc, ωS , at teeth contact. Edges coming
into contact is highlighted in red.
Figure 2.15: Coupling disc teeth as manufactured (bottom) and after severe clashing (top).
Clash plastically deforms and wear the teeth, leading to rounded teeth edges.
There exist other synchronizer failure modes such as hard shifting, i.e. a high
shifting effort required, poor shifting comfort and blockage, i.e. blocked gear en-
gagement after synchronous rotational speed is reached due to cone seizure [76, 79].
While those failure modes also depend on the synchronizer design and the frictional
properties of the synchronizer, they have not been studied in detail in this thesis.
Clash
Full Partial
No synchronization work performed Some synchronization work performed
Loss of
Inadequate pre- External
blocking function, Desynchronization
synchronization disturbances
TI > TC
Too high Wear, Sudden Sudden

oil viscosity changed rotational rotational
surface speed speed
topography, changes in changes in
Mismatch too low µc powertrain powertrain
synchronizer/
manoeuvering Wear, loss
system of contact High drag
pressure, i.e. losses or
no wear gap clutch drag
Too low
spring
system Lubricant
force due to degradation,
e.g. wear too low µc
Overheated
contact
surfaces,
temporarily
too low µc
Figure 2.16: Different causes of clash.
Furthermore, failed engagement due to tooth-on-tooth contact [94] has not been
studied in this thesis.
2.3.10 Synchronizer testing

1
The most common synchronizer test rig is likely the ZF/FZG SSP-180 Synchromesh
Test Rig [45], which is used in e.g. the CEC synchronizer test standard [62]. How-
ever, the test rig used in this thesis is the truck variant of a µ-COMP synchronizer
test rig [95], which is similar to the SSP-180 with respect to test parameters and
procedure. The test rig used in this thesis is a larger version of the rig that was
used in [14, 51], and is schematically shown in Figure 2.17. An electric motor is
used to accelerate the flywheel to its intended speed. When the intended rotational
speed is reached, the motor is turned off. The inner cone of the synchronizer is
fixed to ground, while the latch cone is mounted on the center shaft of the rig. A
force is applied to the shift sleeve by a hydraulic actuator until the shift sleeve has
passed the synchronization position. Lubricating oil is sprayed from the top of the
synchronizer or through the center shaft of the rig. During synchronization, rota-
tional speed, oil temperature, shift sleeve position, applied force and the resulting
torque is measured. Measurement accuracy is presented in paper D. From that
data, the coefficient of friction for a single cone synchronizer can be determined by
TC · sin(α)
µC = (2.6)
FAx · RC
Flywheel
Oil flow
Electric motor
or
Axial force
Synchronizer
Figure 2.17: Schematic drawing of µ-COMP test rig.
2.4 Maneuvering system
In AMT and DCT, the axial force applied to the synchronizer shift sleeve is gen-
erated by an actuator instead of the force the driver applies to the gear lever.
Additionally, the axial force in the splitter and range group for a MT generally
comes from actuators. Heavy duty trucks use pneumatic actuators for several sys-
tems, including gear shift actuators. An example of a maneuvering system for a
synchronizer is shown in Figure 2.18, consisting of a valve unit, a pneumatic actu-
ator (cylinder, piston, position sensor) as well as a gear selector shaft and selector
fork. Main design parameters include cylinder/piston diameter, diameter of air
inlet and system mass.
2.5. OTHER MEANS TO SYNCHRONIZE A GEARBOX AND NOVEL USES
OF SYNCHRONIZERS 31
Pneumatic actuator
Gear shift shaft

Gear shift fork
Synchronizer Valve unit
Figure 2.18: Splitter maneuvering system, Scania’s current (2018) gearbox.
2.5 Other means to synchronize a gearbox and novel uses

of synchronizers
The gearbox can also be synchronized by the engine. Before the synchronizer was
introduced, the driver was forced to “double clutch” [96] (not to be confused with
the transmission type dual clutch transmission, DCT). To synchronize the rota-
tional speed of the shaft and the gear wheel to be engaged in an “unsynchronized”
gearbox, the driver needed to open the clutch, engage neutral gear, close the clutch,
control the engine to the correct rotational speed for the gear to be engaged, open
the clutch, engage the gear, and close the clutch. This process required skill and
timing from the driver, offered poor driving comfort and might lead to long shift
time [55]. The same process is generally used in modern AMT, but is controlled by
the control software instead of the driver [97]. An alternative to opening the clutch
is to reduce the torque from the engine so that the clutch does not transfer any
torque, and then disengage or engage the gear. The downside of this process is that
the engine is slow to retard during upshifts. Systems like an exhaust brake can be
used to increase the retardation performance of the engine, but that reduces the ro-
tational speed of the turbocharger, thus temporarily reducing the engine power and
response after the gear shift. However, for downshifts this process works well since
the engine has good acceleration performance. To improve upshift performance
without the use of synchronizers, a brake unit acting on the lay shaft can be used.
The brake unit is often a pneumatically operated wet multi-disc brake directly con-
nected to the lay shaft, and the synchronizers are replaced by dog clutches. Such
a gearbox is often called “unsynchronized”, which is not strictly correct. It might
lack the traditional cone shaped synchronizers for the main gearbox (however, such
a gearbox generally retain them in the splitter and range), but the main gearbox
gear shifts are synchronized by the brake unit or the engine. Since the brake unit
has similar properties as a synchronizer, i.e. boundary lubricated sliding friction

with rapidly changing relative rotational speed, parts of this research might be
applicable for such a system.
Instead of using a brake unit, an electric motor can be used [98]. With the
increased focus on vehicle hybridization, this solution might become more common
in the future since an electrical motor used to save fuel can also be used to synchro-
nize the gearbox. A novel example of synchronizer usage is for clutchless AMT for
an electric vehicle [60]. Since there is no clutch that can be disengaged, the inertia
to synchronize will be high. Therefore, the speed synchronization is done mostly
by the electric motor, while the synchronizer blocks gear engagement until speeds
are synchronized and then acts as an “engagement unit”.
Another novel use of a synchronizer is the Scania freewheeling retarder [99]. A
retarder is a hydraulic auxiliary brake system where oil is supplied between a stator
and a rotor, and the resulting torque acts to brake the vehicle [100]. However, to
reduce fuel consumption when the retarder is not active, the system is drained of oil
and mechanically disengaged. To mechanically engage the retarder, a synchronizer
is used to accelerate the retarder gear wheel to the speed of the output shaft.
2.6 Lubrication regimes

Many engineering applications require lubrication to function as intended. Lu-
bricants can take many forms, from solid lubricant such as molybdenum disulfide
(MoS2 ), to liquid such as oil or a viscoelastic grease. The lubricant has several
functions, such as:
• Separate surfaces to reduce friction.

• Protect the surfaces by forming tribofilms, avoiding metal-to-metal contact.
• Cooling.
• Protect from corrosion.
• Remove contaminants such as wear debris.
• Dampen noise.
A lubricant consists of a base oil and additives that enhance the base oil prop-
erties [101]. Modern automotive gearbox lubricants are complex products that
need to fulfill many targets. A modern manual transmission fluid thus have several
additives such as [101, 102]:
• Viscosity index improvers to reduce the viscosity-temperature dependence.

• Friction modifiers to reduce friction.
• Anti-wear additives that form a protective layer on the metal surfaces.
2.7. SLIDING CONTACTS AND SURFACE TEMPERATURE 33
• Extreme pressure additives prevent seizure and protects surfaces under ex-
treme pressure conditions.
• Corrosion inhibitors to reduce corrosion.
• Dispersant to keep contaminants and oxide deposits dispersed in the lubricant.
• Additives that protects the lubricant itself such as antioxidants and foam
inhibitors.
• Detergents to neutralize deposits and keep surfaces clean.
Figure 2.19 shows an example of a Stribeck curve [103], which was originally de-
veloped to relate speed and friction in hydrodynamic bearings. The Stribeck curve
describes the relation between the coefficient of friction, µ, and a function of the
dynamic viscosity η, relative rotational speed ω and applied load P . The Stribeck
curve is often used to describe the three traditional lubrication regimes [104]:
• Boundary lubrication (BL), where the asperity contacts carry the load.
• Full film lubrication (FL), where the lubricant film carries the load, i.e. the
surfaces are separated by the lubricant.
• Mixed lubrication (ML), where the load is carried in part by the asperities,
and in part by the lubricant.
Schematic Stribeck curve

µ
BL ML FL
η · ω/P
Figure 2.19: Stribeck curve with the three lubrication regimes.
2.7 Sliding contacts and surface temperature

Looking at a surface from afar, the surface appears smooth. However, at a smaller
scale, surfaces are rough and chaotic in nature [105]. When rough surfaces come
into contact, the load is carried by a fraction of the apparent surface area [106,
107, 108], i.e. the real contact area is a fraction of the apparent contact area. It
is in this real contact area that the frictional work between two sliding surfaces
arises. Most of the frictional work is converted to heat [109, 110]. It is also in
this real contact area that the consequences of friction, e.g. frictional heating, act
[32]. It shall be noted that some of the heat is applied below the surface due to
subsurface plastic deformation. Kennedy found that the vast majority of frictional
energy was dissipated at or within 5 µm of the surface in single pass tests in a
steel/copper contact [111]. However, in this thesis, it is assumed that all frictional
work is converted to heat at the surface. For steel, this assumption has little effect
on the surface temperature, but does change the thermal gradient in the subsurface
material [32]. Due to the local heating, the surface temperature increases locally.
The maximum contact pressure at the asperities on the surface is often as high as
the hardness of the material [106]. Since the local friction power depends on the
local contact pressure, the heat flux and temperature will be high, resulting in a
so-called flash temperature [112]. The flash temperature can exceed 1000 ◦C. The
flash temperature can act over an area with a diameter in the order of 0.1 mm, with
a duration of less then 1 ms† [113, 114]. Since each asperity is small, the amount
of heat in each asperity is low. However, if a large number of asperities in close
proximity come into sliding contact, the temperature in that entire area increases,
i.e. the asperity flash temperature is no longer independent of nearby asperities
[115, 116, 117]. Thermal expansion increases the apparent contact pressure in that
area, which further increases the frictional heating since more asperities come into
contact [115, 118]. Soom et al . described it as “as an average surface temperature
due to accumulated heat from other asperity contacts” [117], while Burton calls
these areas “thermal asperities” [119]. In this thesis, this temperature is referred to
as focal temperature. Barber simulated this self-energizing effect by assuming three
separate areas of contact, where one was initially raised by 1 nm over the other two
contact areas [120]. The initial load at the areas differed by less than 0.1 %. After
a few seconds, the protruding area of contact carried 99.9 % of the load.
In conformal sliding contacts, this localization of the contact pressure and heat
flux can cause the contact surfaces to develop so called “hot spots”, where the focal
temperature is much hotter than the mean temperature of the surface. In this thesis,
focal temperature is used for the surface temperature of a hot spot. The more severe
types of high temperature hot spots (surface temperatures above 750 ◦C) cause
martensite transformation when they are cooled at a sufficient high rate at the end
of synchronization [113]. Martensite has a lower density than the original austenite,
which causes a local protrusion. Martensite is also very hard, thus inhibiting wear
of this locally highly loaded, protruding, area. The scale of these hot spots is large
compared to the surface roughness [107, 113, 121]. This phenomena is referred to
as thermoelastic instability (TEI) [119, 122]. TEI is dominated by long wavelength
† For materials with high thermal conductivity such as steel. CFRP lining flash temperatures
may have longer duration [113]

2.8. CONTACT MODELING 35
perturbations [107], and above a certain critical velocity the contact operates in
an unstable regime [123, 124]. Hot spots are promoted by long contact length
in e.g. disc brakes (where only a segment of the full circle is in contact) or full
ring contact such as synchronizers and clutches [113]. Other factors promoting hot
spots include high stiffness [125, 126], low thermal conductivity [126], high thermal
expansion [126], high friction at high temperature and low wear rate. Anderson
and Knapp states that [113]
“Metal wear from hot spotting is usually quite small in volume, but
quite effective in degrading the function of a clutch or brake assembly.”
Anderson and Knapp also showed examples of different modes of TEI types [113].
Figure 2.20 shows an example from a molybdenum coated synchronizer with both
band and focal type TEI.
(a)
(b)
Figure 2.20: Examples of TEI in synchronizer. (a) Band type in molybdenum surface. (b)
Focal type in steel surface.
2.8 Contact modeling

When surfaces comes into contact, a contact pressure arise to prevent the surfaces
from physically occupying the same space. Ideally, where there is contact, there
is a contact pressure and no gap or overlap between the surfaces, and where there
is a gap, there is no contact pressure. Mathematically, this can be expressed as a

(linear) complementarity problem (LCP) [127, 128, 129, 130]
p = 0 f or hc < 0 (open)
(2.7)
hc = 0 f or p > 0 (closed)
p is the contact pressure, hc is “overclosure”, i.e. interpenetration, between the
surfaces ‡ .
To model contacts numerically, some different concepts are needed, including:
• Contact surface definition. This includes to manually define which surfaces

that should interact with each other, or to let the software track all surfaces
and search for interaction.
• Contact surface discretization. The discretized FE mesh might be, depending

on element order, non-smooth. This leads to intermittent local contact. To
ensure smooth contact, correction factors can be implemented which accounts
for the difference between the initial, smooth 3D geometry and the discretized
FE model [131]. The correction factor is schematically shown in Figure 2.21.
• Contact properties, i.e. contact stiffness, friction and thermal conductivity.
• Contact formulation. In this work, a non-symmetric surface-to-surface formu-

lation is used. The slave surface is constrained to not penetrate the master
surface. The calculated gap/penetration is based on an integral over the
region surrounding the constrained slave node [132, 133]. This leads to a
smoother contact than node-to-node or node-to-surface discretization.
Master surface
Discretized surface
Smooth surface
Correction factor
Slave surface
Nodes
Figure 2.21: Schematic view of geometric correction between discretized FE mesh and
smooth surface.
In this presented work, the penalty method is used to enforce the contact con-
straints. In the penalty method, a penetration, i.e. overlap, of two contacting
‡ Note that h often denote the gap between the surfaces, but here h follows the definition
c
used by the Abaqus theory guide, i.e. interpenetration of surfaces [127].
2.8. CONTACT MODELING 37
bodies is penalized with a force that depends on the amount of penetration that
acts to separate the contacting bodies [134, 135]. This can be seen as virtual springs
between the contacting bodies. The advantages of the penalty method include that
no extra degrees of freedom is required, and it does not risk overconstraining the
model. The drawbacks include that some penetration is required, which in some
cases can change the contact pattern and thus invalidate the results. Additionally,
the contact stiffness, i.e. the relationship between penetration and separating force,
needs to be defined. This user-defined parameter introduces an uncertainty, since
a correct contact stiffness can be hard to determine. It also introduces the possi-
bility to “soften” the contact, thus reducing the severity of the nonlinearity in the
simulation.
Contacting bodies introduce a highly nonlinear relationship between contact
force and displacement [135]. To solve nonlinear problems, the Newton method is
often used [136, 137]. Newton’s method is a root-finding algorithm where
f (x0 )
x1 = x0 − (2.8)
f 0 (x0 )
f (xn )
xn+1 = xn − (2.9)
f 0 (xn )
x0 is an initial guess of the solution for the function f (xn ), x1 is the result of the
first so-called Newton iteration. x1 is then used as a basis for the next iteration.
Analogously, xn is the result of the n-th Newton iteration, and is used as starting
point for the n+1-th iteration. The process is repeated until xn+1 is sufficiently
close to the solution. In some situations, the Newton method will fail to find the
solution, for example if the problem is non-smooth, i.e. a non-differentiable or
discontinuous function [137]. Sudden changes in the contact condition can cause
the Newton method to fail due to the problem being non-smooth. This can be
mitigated by reducing the severity of the nonlinearity. In Abaqus, the contact
stiffness can be defined as linear, based on tabular values, or exponential according
to [127]
p=0 f or hc ≤ −c
p0
1 hc
hc
1
(2.10)
p = (exp(1)−1) c c + 2 exp c +1 − c f or hc > −c
p0 is the pressure at zero gap, i.e. hc = 0, and c is the initial contact distance, i.e.
the minimum gap where p = 0. Abaqus also offers a “hard” contact option, which
sets “the penalty stiffness to 10 times a representative underlying element stiffness”
[134].
Chapter 3
Research methodology
This section contains a summary of the methodology used in this research. Two
different types of numerical simulations were used, namely fluid-structure interac-
tion (2D and 3D) and thermal/thermomechanical simulations (1D (only thermal),
2D and 3D). Additionally, physical tests were performed in a µ-COMP rig as well
as dedicated material test rigs to estimate material properties.
3.1 Fluid-structure interaction model to simulate

pre-synchronization
The blocking of gear engagement during asynchronous rotational speed is a pre-
requisite for successful synchronization. To study the relation between oil film
thickness at end of pre-synchronization and parameters such as spring system force,
maneuvering system behavior, and groove design, two multiphysics fluid-structure
interaction models were developed and simulated with the commercial FE software
Comsol Multiphysics 4.4. One model is axisymmetric, Figure 3.1, and can be used
to assess the effect of cone geometry and maneuvering system. It was assumed
that the space between the cones was initially filled with oil. The simulation was
run until the maneuvering force exceeded the spring system force. The oil film
thickness was evaluated at the end of the simulation. The other model, which is
3D with a simplified geometry, is available in Figure 3.2. It can be used to assess
the effect of different axial grooves in the contact surface. The oil outlets for both
models were modeled by a Dirichlet boundary condition enforcing zero pressure
at the boundary. To avoid negative pressures in the grooves, a cavitation model
where the oil viscosity and density is reduced based on pressure was developed and
implemented. Söderfjäll later implemented the same model, and verified that it
is mass-conservative and yields the same solution as LCP formulation except for
some negative pressures that have to be disregarded [138]. It is significantly easier
to implement and less computationally expensive than the LCP model. To reach
convergence, the 3D model had a lower applied force than the 2D model. Instead
39
40 CHAPTER 3. RESEARCH METHODOLOGY
of the stopping condition used in the 2D model, all 3D simulations were run for one
second. Thorough descriptions of the models are given in paper A.
Out
le
t
F
Oi
l
“
Wal
l” La
tchc
one
Out
le
t
Figure 3.1: 2D Fluid-structure interaction model.
(
a) (
b)
(
c)
Figure 3.2: (a) Synchronizer, (b) simplified 3D fluid-structure interaction model, and (c)
grooves in synchronizer contact surface.
3.2. THERMOMECHANICAL MODEL TO SIMULATE MAIN
SYNCHRONIZATION 41
3.2 Thermomechanical model to simulate main
synchronization
A synchronizer’s behavior during synchronization is a combination of mechanical
and thermal effects. Mechanical effects include the moment of inertia to synchronize
and the contact mechanics between the interacting parts. Thermal effects include
heat generated in the friction surface during synchronization as well as heat flux
in the cones. Thermal expansion deforms the contacting surfaces and localizes
the contact area. A reduced contact area further focuses the heat generation, cre-
ating a self-energizing process [51, 53], also known as a thermoelastic instability,
as described in section 2.7 - Sliding contacts and surface temperature. A multi-
physics thermomechanical model was developed and solved with the commercial
FE tool Abaqus/Standard (initially version 6.13 for papers B and C, later changed
to version 2016). The model consists of two generalized axisymmetric parts (later
converted to a full 3D model, see section 3.6 - Tolerance effects and cone angle
optimization and paper G), as shown in Figure 3.3. The modeling steps are shown
in Figure 3.4.
The model was initially used to perform two parameter studies of a synchro-
nizer. The first study investigated how different external loads affect the tem-
perature increase in the synchronizer. The second study investigated how different
synchronizer geometries distribute the heat in the synchronizer by simulating synch-
ronization processes with the same load, but for different geometries. A thorough
description of the model is available in paper B. Synchronizers with both molybde-
num coating and CFRP lining were investigated. The material properties used are
shown in Table 3.1. At this stage, no experimental material data was available for
the CFRP lining, so most values were taken from literature [14]. It shall be noted
that the Young’s modulus is significantly overestimated for through-plane compres-
sion, and more relevant for the in-plane properties. However, as shown in section
3.2.1 - Contact properties used in this thesis, it had little effect on the contact
pressure distribution since an exponential pressure-overclosure relationship that is
not based on the underlaying material stiffness was used. The CFRP lining has
significantly lower thermal conductivity than molybdenum and steel, which affects
the energy partition between the cones. An implicit heat partition model was used,
where an arbitrary heat partition was applied to the contact, and a very high ther-
mal conductance over the contact ensures thermal continuity, to equalize the local
temperature of the interacting surface patches. This method ensures equal surface
temperature, but only works for continuous contact without local separation from
e.g. cone out-of-roundness.
Contact surfaces
Latch cone
Inner cone
Applied force
DoF 5
Teeth area
DoF 2
DoF 1 Connected to Reference point
Figure 3.3: Geometry and discretization of numerical model.
Fix inner cone. Apply an axial displacement to latch cone to establish

contact.
Disable displacement boundary condition, apply a low force to maintain
contact state.
Apply rotational speed to one or both cones. Increase force, reduce
rotational speed of slave system based on torque and inertia until synch-
ronization is finished. Optional: Apply cooling to external surfaces.
Optional: Separate the cones. Apply cooling to contact surface.
Optional: Cooling
Transfer data to MATLAB for analysis.
Figure 3.4: Thermomechanical model flowchart.

SYNCHRONIZATION 43
Table 3.1: Material parameters for the initial thermomechanical simulation. Data for
CFRP lining is updated for paper F and paper G according to section 3.5 - CFRP material
data and simulations.
Parameter Unit Steel CFRP Molybdenum

3
Density [kg/m ] 7850 1050 10280
Young’s modulus [GPa] 209 50 329
Poisson’s ratio [-] 0.3 0.4 0.31
Coef. of thermal expansion [10−6 /K] 11 70 4.8
Thermal conductivity [W/(m K)] 48 0.65 138
Specific heat capacity [J/(kg K)] 452 1770 250
3.2.1 Contact properties used in this thesis

The following contact stiffness definitions have been used in this thesis:
• For axisymmetric model of molybdenum coated synchronizer, and

CFRP lined synchronizer in paper B and paper C: Exponential contact
stiffness (equation 2.10, section 2.8 - Contact modeling). The values for p0
and c is 104 Pa and 5 × 10−6 m, respectively. This method means that the
contact stiffness is not directly based on the underlying material stiffness.
This showed good correlation with physical testing as shown in paper C.
• For molybdenum coated synchronizer, 3D model: Linear contact stiff-

ness equal to the effective Young’s modulus, E ∗ , according to
1 − ν12 1 − ν22

E∗ = 2 · + (3.1)
E1 E2
where E and ν is the Young’s modulus and Poisson’s ratio for material 1 and 2.
When adding tolerance effects such as cone out-of-roundness, the previously
used exponential contact stiffness became very stiff under some conditions,
especially at higher load. Additionally, the exponential contact stiffness led
to convergence problems since the contact zone is rapidly moving between
iterations, with sudden changes in local contact pressure. A linear contact
stiffness was used instead, which solved the convergence problem. It was
believed that the stiffness was overestimated using the exponential contact
stiffness model, especially since no wear model was applied. The linear contact
stiffness definition yields largely the same results in an axi-symmetric model,
i.e. maximum surface temperature and surface temperature distribution in
the contact surface, as the exponential contact stiffness definition at loads
similar to the loads in paper D. At lower loads, it yields a more focused contact
zone and thus a higher maximum temperature. At higher loads it yields a
less focused contact zone, and therefore lower maximum temperature.
• For CFRP lined synchronizer: Abaqus “hard” contact, i.e. linear contact
stiffness of 10 times a representative underlying element stiffness is used [134].
This allows for relatively small penetration, and worked fine with the CFRP
lining due to its lower Young’s modulus.
Using “hard” contact between two surfaces with high Young’s modulus might lead
to situations where only one or a few nodes carry the load. This contact stiffness
definition is not suitable for molybdenum coated synchronizers.
3.2.2 Verification and validation

The model was verified by reviewing and comparing output data such as cone torque
and energy levels to analytical results for cone torque, rotational speed and brake
energy. There were good agreements between simulated and analytical results, as
shown in Figure 3.5. A study of the discretization error was also performed in paper
C, and it was shown that a sufficiently small discretization in both space and time
was used in the simulation. The model was validated with bulk and surface temper-
ature measurements as well as a qualitative assessment of wear marks compared to
high temperature areas of the synchronizer. Mileti et al. used a similar procedure
[139]. A pyrometer was installed in the inner cone to measure the surface temper-
ature of the latch cone during synchronization. The pyrometer installation can be
seen in Figure 3.6. The emissivity was estimated experimentally by measuring the
surface temperature of a preheated synchronizer with both thermocouples and the
pyrometer. The emissivity used in the analysis was 0.89 for the molybdenum coat-
ing and 0.87 for the carbon lining. Measured surface temperature compared with
simulated temperature can be seen in Figure 3.7. The agreement for molybdenum
coated synchronizers is good. For carbon lined synchronizers, a correction factor
based on the Abbott-Firestone curve∗ was applied due to the large surface features
and low thermal conductivity of the CFRP lining. A thorough description of the
verification and validation procedure, including the carbon lining correction factor,
is available in paper C.
∗ Sometimes referred to as Bearing area curve. Mathematically, its a cumulative density func-
tion of the surface height of a rough surface, showing the fraction of material over each specific
surface height [107]
SYNCHRONIZATION 45
(a) Force and Torque

250
Axial force [N]
2000
Torque [Nm]
200
150
Force Expected
1000 Force Abaqus "RF2" 100
Torque analytical
Torque Abaqus "RM2" 50
0 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Time [s]
(b) Speed and Energy
Accumulated energy [J]

60 800
Speed [RPM]
Speed Expected 600

40 Speed Abaqus
Energy analytical 400
Energy Abaqus
20
200
0 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Time [s]
Figure 3.5: Thermomechanical model verification by comparing analytical estimates to FE

output data.
Figure 3.6: Pyrometer installed in inner cone, mounted in µ-COMP rig. The pyrometer
measures the latch cone surface temperature.
Surface temperature increase, Mo, 600 RPM

Temp. increase [◦ C]
20 Measured
Simulated
10
0
−0.5 0 0.5 1 1.5 2
Time [s]
Surface temperature increase, CFRP, 600 RPM
30 Measured
Measured · 1/0.6
20 Simulated
10
0
−0.5 0 0.5 1 1.5 2
Time [s]
F
Figure 3.7: Validation of thermomechanical model. The correction factor 1/0.6 is briefly
discussed in section 5 - Summary of appended papers and thoroughly discussed in paper
C.
3.2.3 1D thermal model

To remove the need for time-consuming finite element simulations to determine the
average surface temperature, a simplified 1D model was developed in paper D. The
Peclet number can be used to support this simplification. The Peclet number is
defined as
L·v
Pe = (3.2)
4·K
where L is the contact length, v is the sliding speed and K is the thermal diffusivity
of the material. For fast moving contacts such as those in a synchronizer, the
Peclet number is often in the order of 1000. According to Johnsson [107], the
heat conduction may be treated as one-dimensional, i.e. the in-plane heat flow is
insignificant compared to the heat flow from the surface into the material, if the
P e > 5. A Neumann boundary condition (i.e. prescribed heat flux) was applied to
the node representing the contact surface. Since the synchronization time is short,
the model is assumed to be adiabatic. Therefore, a Dirichlet boundary condition
(i.e. prescribed temperature) was applied to the material sufficiently far from the
friction surfaces. A flowchart of the model is available in Figure 3.8. AC is the
SYNCHRONIZATION 47
nominal cone area, ωL is the lowest rotational speed difference to consider, γ is
the heat partition factor between the cones, λ is the thermal conductivity, ts is the
model time step, h is the node to node distance, t is the current simulation time, q̇
is the heat flux density, Tnc ,t is the temperature of node nc at time t. Note that the
model is valid if and only if the time step, ts , equals the critical time step, tCritical
[140]
h2 · ρ · cp
tCritical = (3.3)
2·λ
where ρ is the density, cp is the specific heat capacity. The simplified thermal model
is about four orders of magnitude faster than the full FE model (0.03 s compared
to 300 s). In paper E, it was updated to accept measured data from the µ-COMP
rig directly, but the principle is the same.
Input:
AC , ω L ,
γ, λ, ts , h,
TC (t), ω(t)
No
ω(t) > ωL Synchronized
Yes
t = t + ts
TC (t)·ω(t)
q̇(t) = γ · AC
h
T1,t = q̇(t) · λ + T2,t−ts
for 2 : nD
Tnc ,t = 0.5 · (Tnc −1,t−ts + Tnc +1,t−ts )
Figure 3.8: Flowchart of 1D thermal model.

3.2.4 Integrating simulations and physical tests

The thermomechanical FE model can import measured data from the µ-COMP
test rig, providing a virtual representation of an actual synchronization cycle to
assess surface temperature during synchronization. Applied force and rotational
speed as functions of time is applied as load and boundary condition in the FE
model, while the coefficient of friction as a function of sliding speed is applied
to the contact properties. The resulting torque from the model can be used to
validate the simulation by comparing with measured values, as seen in Figure 3.9.
A thorough description of this methodology is available in paper D.
Measured and simulated torque
120
Cone torque [Nm]
100
80
60
40 Measured
20 Simulated
0
−0.05 0 0.05 0.1 0.15 0.2 0.25 0.3
Time [s]
Figure 3.9: Comparison between measured and simulated cone torque.
3.3 Friction model for main synchronization

The frictional torque between the latch cone and the inner cone is important for
both functionality and performance, as explained in sections 2.3.3 - Preventing un-
synchronized gear engagement and 2.3.5 - Main synchronization phase. To better
understand the behavior of a synchronizer, a friction model was developed. It has
been shown in literature that the coefficient of friction for a given contact interface
depends on sliding speed, contact pressure and temperature [67, 68]. Coefficient of
friction, sliding speed and applied force is measured during synchronizer testing, as
explained in section 2.3.10 - Synchronizer testing. As explained in section 3.2.4 -
Integrating simulations and physical tests, a virtual representation of a test cycle
can be created, and the surface temperature increase during synchronization can
be obtained. Thermocouples were used to measure cycle start temperature, and by
adding starting temperature and temperature increase, transient surface temper-
ature could be obtained. A test cycle with different axial forces at different start
temperatures was developed. An equation to describe the coefficient of friction
as a function of sliding speed, axial force and surface temperature was proposed.
3.4. SPEED STEP TEST TO DETERMINE LIMITING LOAD 49
Residual analysis was used to analyze the accuracy of the proposed equation. This
process was iterated until a satisfactory fit was obtained. The workflow can be
seen in Figure 3.10. A detailed description of the friction model and the simplified
thermal model is available in paper D.
Physical testing Simulation Calculation Output
FAx
µC
Propose
TC equation
FE model Model of
v Ta , pa µ
Evaluate
residuals
Tinitial
Figure 3.10: Friction model development workflow.
3.4 Speed step test to determine limiting load

In order to predict failure in molybdenum coated synchronizers, a speed step test
was run in the µ-COMP rig. After an initial running-in procedure, 1000 cycles were
run with a certain inertia and maximum force. When the 1000 cycles have been
completed, the speed difference to synchronize is increased. In order to get similar
temperature at the start of each cycle, the mean power (including cooling time)
per unit area, Pmean , was set to 20 mW/mm2 . The temperature at the inside of
the inner cone was measured with a thermocouple to verify this assumption. This
gives a cycle time, tC , of
ωI2 · I
tC = (3.4)
2 · AC · Pmean
where ωI is the initial relative rotational speed. When the test was completed,
cycles 50, 150, ..., 950 of the last completed speed step was analyzed by comparing
four different loading parameters:
• Maximum specific synchronization power, i.e. power per (nominal) unit area.
The specific synchronization power, PS , is calculated by PS = TC (t)·ω(t)
AC .
• Specific energy to synchronize, i.e. energy per (nominal) R unit area. The
specific energy to synchronize, ES , is calculated by ES = PS dt.
• Focal surface temperature increase. The focal surface temperature increase

is determined using FE simulation as described in section 3.2.4 - Integrating
simulations and physical tests.
• Average surface temperature increase. The average surface temperature is

determined using the 1D thermal model as described in section 3.2.3 - 1D
thermal model.
Figure 3.11 shows the speed step cycles as well as the cycles to analyze. Full details
are available in paper E.
Speed step test

800
Run-in Load cycles Fail
600
Speed (RPM)
400 Cycles
to analyze
200 680
0 7,250 8,250
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000
Cycle number
Figure 3.11: Speed step test procedure with cycles to analyze as well as the ultimate failure
highlighted.
3.5 CFRP material data and simulations

The material parameters needed for the FE simulations are the Young’s modulus,
coefficient of thermal expansion (CTE), specific heat, thermal conductivity, density
and Poisson’s ratio. The elastic modulus is used to determine the stiffness of the
parts as well as the contact stiffness, and thus the local contact pressure distribution.
The coefficient of thermal expansion (CTE) dynamically affects the shape of the
contact surface and thus the local contact pressure distribution. The specific heat
and density are used to determine the temperature increase from the added heat.
The thermal conductivity determines the heat flow away from the contact surface.
3.5. CFRP MATERIAL DATA AND SIMULATIONS 51
The Poisson’s ratio is used to relate the radial strains to axial and circumferential
strains, but has little impact on the simulations since the strains are small and
the simulations are dominated by the radial behavior. For molybdenum and steel,
reference values are readily available, and the used values is presented in Table 3.1.
For the CFRP lining, these properties need to be estimated experimentally for each
unique lining type. The CFRP linings are transversely isotropic, i.e. the in-plane
and out-of-plane properties differ. The in-plane properties are dominated by the
carbon fiber properties, while the out-of-plane properties are largely defined by the
matrix properties. To reduce the number of tests needed to estimate the material
data, only data for the out-of-plane direction is measured. Since the Peclet number
is significantly higher than 5, this is a reasonable assumption, as stated in secion
3.2.3 - 1D thermal model. Table 3.2 shows what method is used to estimate the
material properties.
Table 3.2: Material parameters needed for simulation and test method to obtain each
parameter
Parameter Test method

Elastic modulus Compression test
CTE DMA
Specific heat DSC
Thermal conductivity LFA
Density Scale and caliper
Poisson’s ratio Reference [14]
The modulus was measured in an Instron 4505 tensile and compression test rig.
The upper tool half is rigid, while the lower tool half is floating to reduce the risk
and effect of parallelism deviations. The tool halves and sample is schematically
shown in Figure 3.12. Samples of approximately 50 mm by 50 mm were used, where
the CFRP lining was laminated between two 2 mm thick steel plates. The CFRP
lining was bonded to the steel plates with the same phenolic resin adhesive film
used in the synchronizer, using the same bonding process and parameters as for
real synchronizer, to include the effect of the adhesive film as well as the bonding
process. The steel plates are considered rigid. An oven was mounted in the test rig
to heat the samples and tool halves. When the oven reached the test temperature of
180 ◦C, the samples were installed and kept in the oven for 2 h to ensure the correct
temperature. Each sample was tested twice with a load ramp from unloaded to
50 kN (0 to 20 MPa nominal contact pressure). The first load application is assumed
to correct the position of the floating tool half, and the second load application is
evaluated. The procedure is similar to the measurements by Ohkawa et al. [64].
The CTE is measured in a dynamic mechanical analysis (DMA) tester, namely
a Netzsch DMA 242 E, schematically shown in Figure 3.13. A 4 mm quartz rod,
with a CTE of 0 µm/(m K) is pressed against the sample with a static force of 0.3 N,
Rigid tool half Steel
CFRP
Self adjusting
tool half Steel
Figure 3.12: Schematic view of compressive modulus measurement.
and the position is recorded. The temperature is increased from 26 ◦C to 200 ◦C at

a rate of 0.5 ◦C/min. To estimate the CTE of the machine itself, a test without the
sample is performed, and the machine CTE is corrected for during measurements.
F
Quartz rod
Lining
Figure 3.13: Schematic view of coefficient of thermal expansion measurement using DMA.
The specific heat was measured by differential scanning calorimetry (DSC) in

a Netzsch DSC 214 Polyma, schematically shown in Figure 3.14. The temperature
of two holders, one reference and one containing a 7.2 mg sample, is placed in a
furnace. The temperature in the furnace is increased by 5 ◦C/min from 40 ◦C to
230 ◦C. The temperature difference between the holders can be used to estimate
the specific heat of the sample [141].
The thermal conductivity was estimated using laser flash analysis (LFA) in
a Netzsch LFA 427, schematically shown in Figure 3.15. Three layers of CFRP
is laminated together using the adhesive used to bond it to the steel cone, to
ensure the sample is opaque to the laser. A 0.8 µs laser pulse heats one side of the
sample, and an infrared detector records the temperature response on the other
Furnace
Heating elements
Reference Sample
holder
Lining
Thermocouples
Figure 3.14: Schematic view of specific heat measurement using DSC.
side. Measurements was performed at 26, 70, 90, 110, 130, 150, 200 and 250 ◦C.
Five laser pulses is performed at every temperature step. The Cape-Lehman +
pulse correction model was used to estimate the thermal diffusivity [142]. The
thermal conductivity, λ, can be estimated by
λ(T ) = K(T ) · ρ(T ) · cp (T ) (3.5)

where K is the thermal diffusivity, ρ is the density, cp is the specific heat and T is
the temperature.
Infrared detector
Heat with laser flash
Figure 3.15: Schematic view of laser flash analysis using LFA.
The density is estimated by weighing 20 strips of CFRP lining, and determining

their size (39.7 mm by 9.7 mm by 0.58 mm, by a digital caliper) on a Mettler AE
260 DeltaRange. The weight is 4.779 g, and the density 1070 kg/m3 . Porosity is
not explicitly modeled in the simulation model. However, the effect of porosity on
density is considered with the material data, i.e. the density is based on nominal
volume. The Poisson’s ratio is assumed to be constant since it has little effect on
the simulation and no measuring method was available to measure the temperature
dependent Poisson’s ratio. The value 0.3 is taken from reference literature for a
similar lining [14].
The experimentally estimated material data is presented in Figure 3.16. Figure
3.16a shows the stress/strain relationship used in the simulation, as well as the raw
measured data. Figure 3.16b shows the CTE for the CFRP lining. There was a
relatively large spread between the four samples, and the CTE seems to decrease
at temperatures above 150 ◦C. Due to the large spread, the simulations was run
with a constant CTE equal to the average of the four measured curves. Figure
3.16c shows the specific heat of the CFRP lining. Only one test was performed to
estimate specific heat, but the results was in line with the expected value, which is
between the specific heat of carbon fibers and phenolic resin [143, 144]. Figure 3.16d
shows the thermal conductivity and the thermal diffusivity. The thermal diffusivity
is not used in the simulation, but was used to estimate the thermal conductivity.
(a) Stress vs strain (b) Coef. of thermal expansion

20 100
CTE [ m/(m K)]

Measured data Measured data
Average data Average data
Stress [MPa]
15 75
10 50
5 25
0 0
0 0.01 0.02 0.03 100 150 200
Strain [-] Temperature [°C]
(c) Specific heat (d) Diffusivity and conductivity
Specific heat [J/(kg K)]
1800
Conductivity [W/(m K)]
0.6
Diffusivity [mm /s]

Conductivity 0.4
2
1600 Diffusivity, interp.
Diffusivity 0.35
0.5
1400 0.3
0.4
1200 0.25
1000 0.3 0.2

0 100 200 300 0 100 200 300
Temperature [°C] Temperature [°C]
Figure 3.16: Thermomechanical properties from the material tests. (a) Compressive
stress/strain relationship. (b) Coefficient of thermal expansion (c) Specific heat. (d) Ther-
mal conductivity and thermal diffusivity.
The validation data presented in Figure 3.7 (and paper C) is used to validate
the new material data. Figure 3.17 shows that the new material data yields similar
results at low load as the old, assumed, material data.
A sensitivity study is also performed, where each material parameter is mul-
tiplied with 0.5, 1 or 2, and all 81 possible combinations is simulated for three
different load cases. The load cases is a maximum force of 2500 N, an inertia to
synchronze of 0.42 kg m2 , and an initial rotational speed of 500, 1000 or 1500 RPM.
The temperature response is evaluated using Analysis of Variance (ANOVA). The
normalized results are presented in Figure 3.18. The compressive modulus has the
largest effect on the focal surface temperature.
Temp. increase [◦ C] Surface temperature increase, Mo, 600 RPM
20 Measured
Simulated
10
0
−0.5 0 0.5 1 1.5 2
Time [s]
Surface temperature increase, CFRP, 600 RPM
30 Measured
Measured · 1/0.6
20 Sim., paper C
10 Sim., new material data
0
−0.5 0 0.5 1 1.5 2
Time [s]
Figure 3.17: Validation of new material data. The correction factor 1/0.6 is briefly dis-
cussed in section 5 - Summary of appended papers and thoroughly discussed in paper C.
ANOVA results
1
500 RPM
Rel. sum of squares of each source [-]
1000 RPM
0.8 1500 RPM
0.6
0.4
0.2
0
nd.
nd.
Cp
Cp
Cp
p
d.
EC
CT
CT
CT
Mo
Co
Co
nd.
d.
CT
nd.
d.
Mo
Co
d.
Mo
Co
Mo
Figure 3.18: Results of material data sensitivity analysis.

A series of component tests was run in the µ-COMP rig, and the focal tem-
perature at selected cycles was simulated to estimate at what simulated focal tem-
perature levels hot spots, accelerated wear and a reduction in coefficient of friction
occurs. The test cycle is shown in Figure 3.19. Four different synchronizers were
run at different speeds, namely 1000, 1150, 1300 and 1450 RPM. The maximum
axial force was 2500 N and the inertia to synchronize 0.42 kg m. The synchronizer
was lubricated with 2.5 l/min fully formulated MTF at 90 ◦C. The mean power
(including cooling time) is set to 20 mW/m. The synchronization position, average
coefficient of friction and simulated focal temperature is compared, and a visual
comparison of the inner cone contact surface reveals at what load levels hot spots
is visible. Full details is available in paper F.
Test cycle
1400
Load cycles
Initial speed [RPM]
1200
1000 Run-in
800 1000 RPM
1150 RPM
600 Measuring cycles 1300 RPM
400 1450 RPM
0 100 200 300 400 500 600 700
Cycle [-]
Figure 3.19: Test cycle in µ-COMP rig for CFRP lined synchronizer.
3.6 Tolerance effects and cone angle optimization

To include roundness deviations from the manufacturing in the model, the axi-
symmetric model was converted to a full 3D model. The roundness deviations were
assumed to be sinusoidal, and was modeled by translating the nodal coordinates
from Cartesian coordinate system to a cylindrical coordinate system. The radial
coordinates were then moved according to equation 3.6.
Ri (θ) = Ri (θ)N om + A/2 · sin (mθ + φ) (3.6)

where Ri (θ) is the radial coordinate of node i at angle θ, Ri (θ)N om is the nominal
coordinate of node i at angle θ, A is the roundness amplitude, m is the out-of-
roundness order and φ is the relative starting position, i.e. phase shift. The relative
cone angle was simulated by changing the latch cone angle.
As described in section 3.2.1 - Contact properties used in this thesis, the con-
tact was modeled using a penalty formulation. For molybdenum synchronizers, the
contact stiffness was modeled as the effective Young’s modulus according to equa-
tion 3.1. For CFRP lined synchronizers, the Abaqus “hard” contact was used. The
heat partitioning factor (HPF) was determined by using the axi-symmetric model
3.6. TOLERANCE EFFECTS AND CONE ANGLE OPTIMIZATION 57
with different HPF, until the surface temperature of the cones were equal. The
HPF for molybdenum coated synchronizer is 52.25 % of the heat to the latch cone,
and 5.1 % heat to the latch cone for CFRP synchronizers. The thermal conductivity
over the contact was [107]
−1
∗ 1 1
λ = + (3.7)
λ1 λ2
where λ∗ is thermal conductivity over the contact, λ is the thermal conductivity of
the respective materials.
The simulated geometric cases are presented in Table 3.3. The latch cone is
assumed to be oval, while the inner cone is assumed to be hexalobe. The shape of
cones is shown in Figure 3.20, magnified 100 times.
Table 3.3: Tolerance cases used in simulation.
Property Value
Latch cone
Cone angle (Mo) [°] 6.2, 6.3, 6.4, 6.5, 6.6, 6.7, 6.8, 6.9, 7.0
Cone angle (CFRP) [°] 7.2, 7.3, 7.4, 7.5, 7.6, 7.7, 7.8, 7.9, 8.0
Roundness, rel. mag. [-] 0, 0.33, 0.67, 1
Roundness, order [-] 2 (oval)
Inner cone
Cone angle (For Mo) [°] 6.5
Cone angle (For CFRP) [°] 7.5
Roundness, rel. mag. [-] 0, 0.33, 0.67, 1
Roundness, order [-] 6 (hexalobe)
The roundness and angle of 2000 inner cones, 2000 molybdenum coated latch
cones and 850 CFRP lined latch cones were measured. It was assumed that a 6.5°
and 7.5° inner cone would have the same roundness and angle distribution since
they have the same manufacturing process and design, except for the cone angle.
The surface temperature of 4 000 000 pairs of molybdenum coated synchronizer
cones and 1 700 000 pairs of CFRP lined synchronizers were evaluated by cubic
interpolation between the simulated geometrical cases. The measured geometrical
deviations is shown in Figure 3.21. Figure 3.21a shows the relative cone angle, here
defined as αLatch − αInner , where αLatch is the latch cone angle and αInner is the
inner cone angle. The relative cone angle follows a normal distribution. Figure
3.21b and Figure 3.21c shows the roundness deviation of the latch cone and inner
cone, respectively. Both roundness deviations follows a lognormal distribution.
To find the optimal cone angle for the measured population of synchronizers, the
nominal cone angle was changed until the lowest population temperature response
was obtained. Full details is available in paper G.
(a) Latch cone
(b) Inner cone
Figure 3.20: Shape of (a) oval Latch cone and (b) hexalobe Inner cone, worst tolerance
case magnified 100 times.
(a) Relative cone angle: Latch cone angle - Inner cone angle
0.25
CFRP
0.2
Probability
0.15
Molybdenum
0.1
0.05
0
Relative cone angle
(b) Latch cone roundness (c) Inner cone roundness

0.25 0.25
0.2 0.2
Probability
Probability
0.15 0.15
0.1 0.1
0.05 0.05
0 0
Magnitude Magnitude
Figure 3.21: Measured tolerance distribution. Blue data represents molybdenum coated
synchronizers, and red data represents CFRP lined synchronizers. The cone angle for
molybdenum coated synchronizers is 6.5◦ , and 7.5◦ for CFRP lined synchronizers. Only
the 6.5◦ inner cone is measured, and it’s assumed that the tolerance distribution is equal
for a 6.5◦ and a 7.5◦ cone. The values on the X-axis is removed due to confidentiality.
Chapter 4
Results
4.1 Pre-synchronization simulation

Figure 4.1 and Figure 4.2 show the results from pre-synchronization simulations
from the 2D and 3D model, respectively. It can be seen that parameters connected
to the maneuvering system, i.e. force ramp speed and maneuvering system mass,
have a large effect on the pre-synchronization behavior. No significant benefit from
increasing the spring force beyond ~130 N can be seen in this specific system with
these oil properties, but it would be helpful in e.g. cold conditions, where the oil
viscosity is higher. Grooves in the contact surface are highly beneficial, and the
presence of grooves seems more important than groove design.
59
60 CHAPTER 4. RESULTS
(a) Force ramp

150
Oil film thickness [µm]
3500 N/s
7000 N/s
100 14000 N/s
50
0
0 1 2 3 4
Time [s] ·10−2
(b) Pre-synchronisation force
150
100
Oil film thickness over time
50 Oil film thickness at 100 N
Oil film thickness at 130 N
Oil film thickness at 160 N
0
0 1 2 3
Time [s] ·10−2
(c) Actuator mass
150
100
50 0 kg added mass
4 kg added mass
8 kg added mass
0
0 1 2 3
Time [s] ·10 −2
Figure 4.1: Results from 2D pre-synchronization model. (a) Results of different force
ramps. (b) Results of different pre-synchronization forces. A higher spring stopping force
can prevent gear engagement until higher maneuvering forces are reached, resulting in a
longer time to evacuate the oil. (c) Results with different maneuvering system mass.
4.1. PRE-SYNCHRONIZATION SIMULATION 61
(a) Groove width and distance

200
No groove
150 10% groove, 0.5 of 5 mm
20% groove, 1 of 5 mm
100 10% groove, 1 of 10 mm
20% groove, 2 of 10 mm
50
0
0 0.2 0.4 0.6 0.8 1
Time [s]
(b) Groove height
200
No groove
150 0.3 mm
0.6 mm
100
50
0
0 0.2 0.4 0.6 0.8 1
Time [s]
(c) One or two outlets
200
No groove
150 Two outlets
One outlet
100 One outlet, groove depth 8 of 12 mm
50
0
0 0.2 0.4 0.6 0.8 1
Time [s]
Figure 4.2: Results from 3D pre-synchronization model. Example of grooves is available in

Figure 3.2. (a) Groove width and groove distance, assuming either 10 percent or 20 percent
of the apparent contact surface is a groove. (b) Different groove heights. (c) Different
groove depths and number of outlets.
4.2 Main synchronization simulation

Figure 4.3 shows the temperature distribution in a molybdenum synchronizer. It
is clearly shown that the heat flux has been focused to subset of the contact sur-
face, creating a temperature gradient in the axial direction. Figure 4.4 shows the
results from a study of the main synchronization external load parameters. It can
be seen that the rotational speed is the most important parameter, showing an
exponential temperature increase with increasing rotational speed. An increased
moment of inertia increases synchronizer temperature approximately linearly, while
the temperature increase caused by an increased axial force is logarithmic. Figure
4.5 shows the results from the geometrical parameter study. The most important
parameters relate to the contact surfaces, and the bulk geometry only has a small
effect on the surface temperature.
NT11
15
13
11
9
7
5 Latch cone
3
1
0
Inner
cone
Figure 4.3: Schematic temperature distribution in molybdenum synchronizer.
Maximum temperature increase

200
CFRP, Speed
Temperature increase [◦ C]
Mo, Speed
150 CFRP, Force
Mo, Force
CFRP, Inertia
100 Mo, Inertia
50
0
0.5 1 1.5 2
Normalized load [unit]
Figure 4.4: Results from the external load parameter study.

4.3. FRICTION MODEL FOR MAIN SYNCHRONIZATION 63
(a) CFRP (b) Molybdenum
CG:
BD: 600 RPM
BC: 900 RPM
AC: 1200 RPM
G: Teeth pos. (STD)
F: Flange (With)
E: Lining pos. (Latch cone)
D: Stiffness (100%)
C: Straightness (0.025mm)
B: Angle (-0.1◦ )
A: Angle (+0.1◦ )
0 50 100 0 10 20 30
Temp. increase [◦ C] Temp. increase [◦ C]
Figure 4.5: Results from the geometrical parameter study. Category A-C relates to the
contact surface, i.e. angle difference between latch cone and inner cone contact surface as
well as straightness deviation on the latch cone. Category D-G relates to the bulk geometry,
with D being the Young’s modulus of the latch cone, E is the lining position (Latch cone or
inner cone), and F-G are geometrical properties of the latch cone. Full details are available
in paper B.
4.3 Friction model for main synchronization

The coefficient of friction for a molybdenum coated synchronizer can, according to
paper D, be described as:
µC = c1 + c2 · v + c3 · TS + c4 · FAx + (c5 + c6 · FAx ) · tanh (c7 · v) + c8 · ev (4.1)
where µC is the coefficient of friction beween the cones, v is the sliding speed, TS
is the average surface temperature, FAx is the applied axial force and c1 to c8 are
coefficients. Note that Eq 4.1 is only applicable for the material, lubricant com-
bination and surface texture combination tested since it is based on the physical
testing for this combination. A similar sliding speed dependence of the coefficient
of friction is presented in [67]. The values for the coefficients ci are available in
paper D. Figure 4.6 shows a comparison between the measured coefficient of fric-
tion and the modeled coefficient of friction for three different axial forces at three
different temperatures. This data was not a part of the analysis, and only used
for verification. Figure 4.7 shows a comparison between measured and calculated
coefficient of friction for all data points included in the analysis. There is good
agreement between measurements and calculations, except at low sliding speeds,
i.e. close to the gear engagement. Figure 4.8 shows a comparison between the sim-
ulated temperature from the 1D model and the FE model, as well as a comparison
between the measured coefficient of friction and the friction model prediction based
on temperature data from the 1D thermal model for both a single and double cone
synchronizer.
Coefficient of friction
0.12
1000 [N]
0.1
Measured
µ
0.08
Model
0.06
0.12
FAx : 2500 [N] 2000 [N]
0.1
µ
0.08
0.06
0.12
0.1
µ
0.08
0.06
0 0.1 0.2 0.3 0 0.1 0.2 0.3 0 0.1 0.2 0.3
Time [s] Time [s] Time [s]
Oil temp.: 35 [◦ C] 85 [◦ C] 115 [◦ C]
Figure 4.6: Measured coefficient of friction and modeled coefficient of friction for different
maximum forces and oil temperatures.
4.3. FRICTION MODEL FOR MAIN SYNCHRONIZATION 65
Measured and predicted coefficient of friction

0.13
±10% error lines

5
95% of all data points
0.11
4
Sliding speed [m/s]

Calculated µ
0.09 3
2
0.07
0.05
0.05 0.07 0.09 0.11 0.13
Measured µ
Figure 4.7: Comparison between measured and calculated coefficient of friction for all data
points. The majority of modeled data points show good correlation with the measured data.
increase [K]
(a) Single cone
0.12 25
20
0.1
15
µ
Temperature increase Temperature
0.08 10
5
0.06 0
0 µ 1D model
0.05 0.1 0.15
[K]
µ measured Time [s] cone

(b) Double
Temp. 1D model
0.12 Temp. FE 25
20
0.1
15
µ
0.08 10
5
0.06 0
0 0.05 0.1 0.15 0.2 0.25
Time [s]
Figure 4.8: Comparison of simulated temperature increase between the 1D thermal model
and the FE thermal model, as well as a comparison between the measured coefficient of
friction and the modelled coefficient of friction using the 1D thermal model as source of
temperature data for (a) Single cone synchronizer, (b) Double cone synchronizer, outer
contact interface.
Table 4.1: Rotational speed, mean synchronization time and start of cycle temperature
at last completed speed step. *The thermocouple failed for the second test at 3000 N,
0.24 kg m2 . The temperature is assumed to be equal to the first test.
Force Inertia Last completed Mean synch- Start of cycle

[N] [kg m2 ] speed step [RPM] ronization time [s] temperature [◦C]
1000 0.24 1300 / 1300 0.55 / 0.53 100 / 100
1000 0.98 910 / 970 1.38 / 1.62 96 / 98
2000 0.24 1170 / 1250 0.31 / 0.33 99 / 98
2000 0.98 800 / 780 0.68 / 0.65 98 / 98
3000 0.24 1140 / 1110 0.24 / 0.23 92 / (92)*
3000 0.98 690 / 730 0.41 / 0.44 99 / 98
4.4 Molybdenum coated synchronizer load limit

Table 4.1 shows the results from the speed step test with a molybdenum coated
synchronizer. The rotational speed at the last completed speed step is similar
between the repeated tests. The large difference in synchronization time between
the different test cases shows that a large difference in synchronization load is
tested. The start of cycle temperature at the inside of the inner cone shows that
the assumption of equal mean power per area including cooling time gives relatively
equal start of cycle temperature, except for the case of 3000 N and 0.98 kg m2 .
Figure 4.9 shows the focal surface temperature increase from the FE model and
the average surface temperature increase from the 1D model at the last completed
speed step. The surface temperature at the onset of failure depends on the applied
force, and at lower forces also on the inertia to synchronize. At 3000 N, the spread
between the different cycles is relatively large, which is not seen at lower forces.
The average surface temperature increase follows the same trend as the focal surface
temperature increase, but is naturally lower.
Figure 4.10a shows the results of 11 different speed step tests with a force of
2500 N and an inertia of 0.42 kg m2 . The variance in the speed at the last completed
speed step is low. The dashed line represents 920 RPM. Figure 4.10b shows the
result of a 200 000 cycle endurance test at 920 RPM, with a force of 2500 N and an
inertia of 0.42 kg m2 . This shows that there exists a threshold which determines if
the synchronizer fails after just a few cycles, or has a very long service life. The
synchronization time is stable after the initial 30 000 cycles. The indicated wear,
based on synchronizer axial position during synchronization and recalculated to
the cones normal direction, rapidly increases during the first 1000 cycles, and then
decreases before it reaches the expected profile after about 15 000 cycles. The initial
indicated wear is likely due to plastic deformation, while the recovery is likely due
to thermal expansion due to higher mean power in the test compared to the run-in
cycles. Full details is available in paper E.
4.4. MOLYBDENUM COATED SYNCHRONIZER LOAD LIMIT 67
Focal temperature increase, FE model and 1D model

45
2
1000 N, 0.24 kg m
1000 N, 0.98 kg m2
40 2000 N, 0.24 kg m2
2
Temperature increase [°C]
2000 N, 0.98 kg m
3000 N, 0.24 kg m2
2
35 3000 N, 0.98 kg m
FE
1D
30
25
20
Inertia [kg m2] 0.24 0.98 0.24 0.98 0.24 0.98
Force [N] 1000 2000 3000
Figure 4.9: Focal surface temperature increase from FE model and 1D model at last com-
pleted load step for six different load cases.
(a)
Speed step test
1000
Speed [RPM]
950
900
850
800
0 2 4 6 8 10 12
Test number
(b) Endurance test
Synchronization time [s]
300 120
100
250
Wear [µm]
80
200 60
40
150
20
100 0
0 0.5 1 1.5 2
Cycle number ×105
Figure 4.10: (a) Speed step test results, last completed speed step, 2500 N, 0.42 kg m2 .
Dashed line at 920 RPM. (b) Endurance test, 920 RPM, 2500 N, 0.42 kg m2 .
4.5 Results of CFRP simulation and testing

Figure 4.11a shows the measured synchronization position, which is an indication
of wear, as well as the average coefficient of friction for the four different speeds
tested. Figure 4.11b shows the simulated focal temperature at some selected cycles.
The tests at 1000, 1150 and 1300 RPM shows little wear and changes in coefficient
of friction. At 1450 RPM, there is a high wear rate after about 5000 cycles, and a
slight reduction in coefficient of friction. Figure 4.12 shows the inner cone contact
surface after the test. For the 1000 RPM test case, no hot spots is visible. For the
1150 RPM test case, a slight color change is barely visible, which might indicate the
onset of hot spots. At 1300 RPM and 1450 RPM, a “band” type hot spot is clearly
visible [113].
(a) Synchronization position and CoF

Rel. synch. position [mm]
0 0.2
0.18
-0.25
CoF [-]
0.16
-0.5 0.14 Position, 1000 RPM
-0.75 0.12 Position, 1150 RPM
0.1 Position, 1300 RPM
-1 Position, 1450 RPM
0 2000 4000 6000 8000 10000
CoF, 1000 RPM
Cycle [-] CoF, 1150 RPM
(b) Simulated focal temperature CoF, 1300 RPM
CoF, 1450 RPM
260
Temperature [°C]
Temp. 1000 RPM

240 Temp. 1150 RPM
220 Temp. 1300 RPM
200 Temp. 1450 RPM
180
160
0 2000 4000 6000 8000 10000
Cycle [-]
Figure 4.11: (a) Measured results of physical testing: Synchronization position and co-
efficient of friction. (b) Distribution of simulated surface temperature for each initial
rotational speed.
4.6. EFFECT OF MANUFACTURING TOLERANCES AND OPTIMAL
CONE ANGLE 69
Figure 4.12: Inner cone surface after tests. (a) 1000 RPM. (b) 1150 RPM. (c) 1300
RPM. (d) 1450 RPM. At higher speed, the "band" type of hot spot is clearly visible [113].
4.6 Effect of manufacturing tolerances and optimal cone

angle
Figure 4.13 and Figure 4.14 show the maximum simulated focal temperature for
molybdenum coated and CFRP lined synchronizer, respectively. Both cases use the
nominal cone angle, i.e. 6.5° for molybdenum coated latch cone and 7.5° for CFRP
lined latch cone. Note that for molybdenum coated synchronizer, the temperature
increase is shown since no temperature dependent effects exists in the model. For
CFRP synchronizers, the initial temperature is 90°, i.e. same as the oil temperature.
It is shown that roundness deviations always increase the maximum focal surface
temperature, and that the roundness deviations of one cone has little effect on the
temperature of the other cone.
Figure 4.15 and Figure 4.16 show the simulated focal latch cone temperature
for molybdenum coated and CFRP lined synchronizer, respectively, for different
relative cone angles. The different synchronizers react differently to the relative
cone angle, where the 0.5° case is generally the worst for a molybdenum coated
synchronizer, while the −0.3° case is clearly the worst for the CFRP lined synchro-
nizer. CFRP lined synchronizers are thus more sensitive to negative (as defined in
this thesis) relative cone angles than molybdenum coated synchronizers.
Figure 4.17 and Figure 4.18 show the population focal temperature based on
the results in Figure 4.15 and Figure 4.16 as well as the measured manufacturing
deviations. The black line shows the corresponding results from the axi-symmetric
model (i.e. no roundness deviations) with no relative cone angle for different speeds.
The results of a “virtual” population is also included, with an optimized latch
cone with a cone angle of 6.58° for molybdenum coated and 7.61° for CFRP lined
synchronizer. For molybdenum coated synchronizers, there is a relatively large
decrease in maximum focal surface temperature for most synchronizer cone pairs.
However, there still exist a fraction of synchronizer cone pairs with as high focal
surface temperature as for the 6.5° case. For CFRP synchronizers, there is an
even larger reduction in maximum focal surface temperature, and even the worst
tolerance cases are significantly reduced.
CONE ANGLE 71
Mo. Temperature increase for latch cone and inner cone
80
Maximum cone temperature increase [°C] Inner cone
70 Latch cone
60
50
40
30
20
10
0
1 0.67 0.33 00 0.33 0.67 1
Rel. inner cone roundness
Rel. latch cone roundness
Figure 4.13: Molybdenum coated latch cone (6.5°) and inner cone temperature increase for
different levels of out-of-roundness.
CFRP. Temperature increase for latch cone and inner cone

350
Inner cone
Maximum cone temperature increase [°C]
Latch cone
300
250
200
150
100
1 0.67 0.33 00 0.33 0.67 1

Figure 4.14: CFRP lined latch cone (7.5°) and inner cone temperature increase for differ-
ent levels of out-of-roundness.
Mo. Cone angle difference: Latch cone angle - inner cone angle
Maximum latch cone temperature increase [°C] 80 0.5

70 0.4
60 0.3
50 0.2
40 0.1
30 0
20 -0.1
10 -0.2
0 -0.3
0 1
0.33 0.67
0.67 0.33
10
Figure 4.15: Molybdenum coated latch cone temperature for different relative cone angles
and levels of out-of-roundness.
CFRP. Cone angle difference: Latch cone angle - inner cone angle
350 0.5
Maximum latch cone temperature [°C]
0.4
300
0.3
250 0.2
0.1
200
0
150 -0.1
-0.2
100
-0.3
0 1
0.33 0.67
0.67 0.33
10
Figure 4.16: CFRP lined latch cone temperature for different relative cone angles and
levels of out-of-roundness.
CONE ANGLE 73
Mo. Population temperature

Mo. Initial speed difference [RPM]
600 800 1000 1200 1400 1600
80
Maximum latch cone temperature increase [°C]
70 Corresponding speed
assuming nominal geometry
60
50
40
30
20
10 Temperature increase at used load 6.5°

assuming nominal geometry 6.58°
0
0 0.02 0.04 0.06 0.08 0 0.001
Probability
Figure 4.17: Population temperature increase for 6.5° and 6.58° molybdenum coated latch
cone.
CFRP. Population temperature

CFRP. Initial speed difference [RPM]
600 800 1000 1200 1400 1600
300
Corresponding speed
Maximum latch cone temperature [°C]
assuming nominal geometry

250
200
150
7.5°
Temperature at used load 7.61°
100 assuming nominal geometry
0 0.02 0.04 0.06 0.08 0.1 0.12 0 0.001
Probability
Figure 4.18: Population temperature increase for 7.5° and 7.61° CFRP lined latch cone.
Chapter 5
Summary of appended papers
This thesis consists of a summary and seven appended papers. In paper A, the
transition between the pre-synchronization phase and the main synchronization
phase is investigated. In paper B and C, a thermomechanical simulation model
is created, verified and validated. In paper D, the thermomechanical simulation
model is combined with physical testing to develop a friction model. In paper E, the
maximum load a molybdenum coated synchronizer can handle is identified using a
combination of simulations and physical testing. In paper F, material properties
for CFRP lining are measured, and used to identify when hot spots, reduced co-
efficient of friction and severe wear start. In paper G, the effect of manufacturing
tolerances is investigated, and an optimum relative cone angle is identified.
Paper A: Robust pre-synchronization in heavy truck transmissions

The pre-synchronization phase was studied by simulating the evacuation of lubri-
cating oil from the contact surfaces. Two models were used.
• A 2D axisymmetric model that can be used to investigate the interaction be-
tween the maneuvering system (actuator/driver behavior, gear selector shaft
and gear selector fork) and contact surface shape and parallelism. The out-
put of the simulation is the oil film thickness at the transition between the
pre-synchronization phase and the main synchronization phase.
• A 3D model with simplified geometry that can be used to investigate the
addition of grooves in the contact surface.
It was shown that the pre-synchronization is highly dependent on the maneuvering
system, and that the addition of grooves is beneficial.
Paper B: Parameter study of the thermomechanical performance of

heavy duty synchronizers
A thermomechanical FE model to simulate the combined mechanical and thermal
load during synchronization was developed. Modeling complications were discussed,
75
76 CHAPTER 5. SUMMARY OF APPENDED PAPERS
such as how to define the contact stiffness and how to model the rotational speed
using an analytical procedure that does not support moment of inertia. A simple
method for implicit energy partition between the two cones was presented. Both
molybdenum coated and CFRP lined synchronizer were considered.
Two parameter studies were performed. The first parameter study investigated
the effect of different applied synchronizer loads, and showed that the rotational
speed to synchronize has a larger influence than the moment of inertia to synchro-
nize and the axial force. The second parameter study investigated the effect of the
cone geometry, and showed that the cone surface modification has a significantly
larger influence over the resulting temperature increase than the bulk geometry of
the synchronizer.
It was also shown that molybdenum coated synchronizers show similar thermal
trends as a CFRP lined synchronizer, but the surface temperature was significantly
higher for the CFRP lined synchronizer due to the low thermal conductivity of the
CFRP lining.
Paper C: A verified and validated model for simulation-driven design

of heavy duty truck synchronizers
A methodology for verification and validation of thermomechanical synchronizer
simulations was presented. The numerical verification was described and the results
showed that there was good agreement between analytical values and simulated re-
sults. It was also shown that the discretization error was negligible.
To validate the model, both bulk and surface temperature measurements were
performed. Bulk temperature measurements with thermocouples at three different
positions showed good agreement with simulated data just after synchronization,
but no temperature peak during synchronization was visible. This is not intu-
itive, and it was assumed to be the result of a high thermal impedance in the
interface between the thermocouple and bulk material. However, the test vali-
dated the energy partition between the contacting parts. The surface temperature
during synchronization was measured with a pyrometer. The emissivity was deter-
mined experimentally. Molybdenum coated synchronizers showed good agreement
between simulated and measured values. Simulated temperature values for CFRP
lined synchronizers were significantly higher than measured values, and a correction
factor based on the Abbott-Firestone curve was proposed and discussed. With the
correction factor, there was good agreement between measured and simulated data
for CFRP lined synchronizers.
Paper D: Predicting friction in synchronizer systems

A methodology for developing friction models for synchronizers was proposed, and
exemplified by a molybdenum coated synchronizer. It was shown that the coefficient
of friction, sliding speed and surface temperature have strong mutual dependences.
A test procedure was developed for the µ-COMP test rig to test different com-
binations of applied axial force and temperature. Measured data for axial force,
coefficient of friction and rotational speed were imported into the thermomechanical
77
model developed in paper B to virtually represent each test cycle to simulate the
temperature during real test cycles. The simulated torque corresponded well with
the measured torque. The combination of physical testing and simulation yielded
about 50 000 data points containing coefficient of friction, sliding speed, axial force
and surface temperature. Axial force was assumed to be proportional to contact
pressure on the nominally smooth surfaces.
An empirical description of the coefficient of friction as a function of sliding
speed, axial force and surface temperature was developed by least square analysis.
The model showed good correlation to measured data. Additionally, a simplified
thermal model was developed to allow for quick assessment of surface temperature
as well as coefficient of friction. The model was solved iteratively due to the strong
dependence between coefficient of friction, sliding speed and temperature. The
thermal results corresponded well with results from the FE model, which means
the coefficient of friction results also corresponded well with measured data. The
simplified method is about four orders of magnitude faster than the full FE model
(0.03 s compared to 300 s).
Paper E: Evaluation of synchronizer loading parameters and their ability

to predict failure
Molybdenum coated synchronizers were tested in a µ-COMP rig with stepwise in-
creasing rotational speed difference until failure. Axial forces from 1000 N to 3000 N
were tested with an inertia of 0.24 kg m2 as well as 0.98 kg m2 . Four different load-
ing parameters were used to analyze data for 10 synchronization cycles at the last
completed speed step. The analyzed loading parameters was:
• Specific energy to synchronize.
• Maximum specific synchronization power.
• Average surface temperature increase.
• Focal surface temperature increase.
The test rig measurements were used to define the simulation cases for the two tem-
perature models, a procedure developed in paper D. It was shown that both the
average and focal surface temperature increase predict failure with relatively good
accuracy. Additionally, it was shown that there exists a threshold that determines
if the synchronizers fails after a few cycles, or have a very long service life.
Paper F: Thermomechanical performance of CFRP synchronizer fric-

tion liners
Material data for thermomechanical simulations was experimentally estimated for
a CFRP lining. Compressive testing at elevated temperature was used to estimate
the compressive modulus for the lining. Dynamic mechanical analysis was used to
estimate the coefficient of thermal expansion. Differential Scanning Calorimetry
78 CHAPTER 5. SUMMARY OF APPENDED PAPERS
was used to estimate the specific heat. Laser Flash Analysis was used to estimate
thermal conductivity. A sensitivity analysis was performed, and it was shown that
the compressive modulus is the most important material parameter. By simulating
test cases from a physical test, it was found that hot spots started to appear at a
simulated focal temperature of around 200 ◦C, while severe wear and a reduction
of coefficient of friction start at 230 ◦C - 250 ◦C.
Paper G: Optimization of synchronizer cone angle with regards to man-

ufacturing tolerances of cone roundness and cone angle
The axisymmetric thermomechanical simulation model was converted to a full 3D
model. Geometrical deviations for roundness and cone angle were introduced to
represent manufacturing tolerances. The roundness deviations were assumed to be
sinusoidal, with different amplitude and frequency, i.e. shape, for the latch cone
and the inner cone due to their manufacturing processes. When both cones had
roundness deviations, the relative starting position, i.e. phase shift, is important to
find the maximum temperature during synchronization. To reduce the number of
simulations needed, only the latch cone temperature was evaluated. The roundness
and cone angle of synchronizer cones were measured and cone pairs were formed.
4 000 000 molybdenum synchronizer cone pairs and 1 700 000 CFRP lined synchro-
nizer pairs were evaluated. The simulated maximum focal temperature of these
cone pairs was evaluated. A relative cone angle was introduced by changing the
nominal cone angle, and optimum nominal cone angles for molybdenum coated and
CFRP lined synchronizers was determined, which could reduce the focal surface
temperature of the measured synchronizer population.
Chapter 6
Conclusions, discussion and future

work
6.1 Conclusions
The answers to the research sub-questions is:
RQ1: How can the frictional behavior at transition between pre-synchronization

and main synchronization be predicted?
In paper A, the oil film thickness at the transition from the pre-synchronization
phase to the main synchronization phase is assessed. The oil film thickness is an
indicator of which lubrication regime the system is working in. The simulated oil
film thickness is in many cases smaller than the asperity heights, which would in-
dicate boundary lubrication and successful pre-synchronization. The effect of axial
grooves in the contact surface is evaluated, and it is shown that the grooves signif-
icantly aid the oil evacuation during the pre-synchronization phase.
RQ2: How does the coefficient of friction depend on the operating conditions during
main synchronization, and how can it be modeled to allow for more detailed systems
simulations?
A friction model was developed and presented in paper D. The model represents the
coefficient of friction well during main synchronization. The coefficient of friction
depends on the sliding speed, the applied axial force and the contact temperature.
Above 0.5 m/s sliding speed, the coefficient of friction of one cycle increases slightly
with a reduced sliding speed. Around 0.5 m/s sliding speed, the coefficient of fric-
tion decreases rapidly. For a synchronizer, this means that the gear engagement
starts around this point. A 1D thermal model was developed, to efficiently predict
the coefficient of friction, allowing the friction model to be implemented in systems
79
80 CHAPTER 6. CONCLUSIONS, DISCUSSION AND FUTURE WORK
simulation of gear shifts and synchronizers. The 1D model removes the need for
full FE thermal simulations to estimate the average surface temperature, which is
used in the friction model.
RQ3: How do external loads as well as different characteristic design parameters

and their interactions affect the nominal and local contact interface temperatures
during synchronization?
How the contact surface temperature increase depends on three loading param-
eters as well as on geometrical parameters is presented in paper B. Increasing the
rotational speed to synchronize increases the temperature exponentially. Increas-
ing the moment of inertia to synchronize increases the temperature linearly. An
increase in maximum axial force increases the temperature logarithmically. The
straightness and parallelism of the contact surfaces is significantly more important
than the bulk geometry of the synchronizer. The interaction effects between geo-
metrical parameters are low, except for parameters related to the contact surface
geometry. CFRP lined and molybdenum coated synchronizers behave similarly,
but the temperature increase in a CFRP lined synchronizer is significantly higher
than in a molybdenum coated synchronizer due to low thermal conductivity of the
CFRP lining, which affects the heat partitioning between the cones.
RQ4: How does the service life of a molybdenum coated synchronizer depend on
the applied load, and how can the engineers developing synchronizers ensure that
the synchronizer has sufficient service life?
In paper E, it is shown that the service life of a molybdenum coated synchronizer

strongly depends on the applied load. There is a temperature increase threshold
which determines if the synchronizer fails after just a few cycles, or has a very long
service life. Both the simulated average surface temperature increase and focal sur-
face temperature increase is shown to predict failure relatively well. The threshold
for failure depends on the applied force, and shows a weak dependence on inertia to
synchronize at lower forces. Traditional loading parameters such as specific energy
to synchronize and maximum specific synchronization power show poor correlation
with failure.
RQ5: How can the material properties of a CFRP friction lining be estimated,
and how do they influence the surface temperature?
The material properties are estimated in paper F. The compressive modulus can
be estimated using compressive testing, preferably at elevated temperatures. The
coefficient of thermal expansion can be estimated by dynamic mechanical analysis
(DMA). The specific heat can be estimated by differential scanning calorimetry
(DSC). The thermal conductivity can be estimated by laser flash analysis (LFA). It
is shown that the compressive modulus has the largest influence on the focal surface
6.1. CONCLUSIONS 81
temperature. The thermal conductivity has a relatively large effect at lower loads,
while the coefficient of thermal expansion becomes more important at higher loads.
The influence of the specific heat is relatively important in all considered cases.
RQ6: Is there a surface temperature in a synchronizer with CFRP lining, above

which surface damages can be observed and accelerated wear and performance degra-
dation will occur?
For the tested cases, hot spots did not appear when using an initial speed of
1000 RPM and clearly appeared when using an initial speed of 1300 RPM. Signs
of hot spots was barely visible when using an initial speed of 1150 RPM. The
corresponding simulated focal surface temperature is 180 ◦C for 1000 RPM, 200 ◦C
for 1150 RPM and 230 ◦C for 1300 RPM. The transition between mild and severe
wear is between the 1300 RPM and 1450 RPM load case, which corresponds to a
simulated focal surface temperature between 230 ◦C and 250 ◦C.
RQ7: How do manufacturing tolerances affect the synchronizers maximum sur-

face temperature during synchronization? Can the effects be reduced by changing
the nominal design?
The effect of manufacturing tolerances on the focal surface temperature was evalu-
ated in paper G. Cone out-of-roundness always increases the maximum focal surface
temperature. The latch cone out-of-roundness has little effect on the inner cone
focal surface temperature, and the inner cone out-of-roundness has little effect on
the latch cone focal surface temperature. It is thus not suitable to increase the tol-
erance width of one cone just because it is possible to tighten the tolerance on the
other cone. A relative cone angle can have a large effect on the focal surface tem-
perature. CFRP lined synchronizers are more sensitive to “negative” relative angles
than molybdenum synchronizers. Based on the measured geometric deviations, the
optimum nominal relative cone angle is 0.08° for molybdenum synchronizers and
0.11° for CFRP lined synchronizers. For molybdenum synchronizers, the highest
focal surface temperature will be as high for a design with no nominal relative cone
angle as for a design with a nominal relative cone angle of 0.08°, but the highest
temperature will occur less frequently. For CFRP lined synchronizers, the whole
population shows a significant reduction in focal surface temperature when a nom-
inal relative cone angle is introduced, especially for the tolerance cases that had a
high focal surface temperature with no nominal relative cone angle.
RQ8: How can thermomechanical synchronization models be validated?
In paper C, a thermomechanical synchronizer model is validated with bulk temper-

ature measurements using thermocouples and surface temperature measurements
using a pyrometer. For the surface temperature measurements, the emissivity was
estimated experimentally. The reported temperature was an average of a circular
82 CHAPTER 6. CONCLUSIONS, DISCUSSION AND FUTURE WORK
area with a diameter of 7 mm. The measurements validated the average temper-
ature over the measured area, as well as the time constant of the temperature
increase. A qualitative validation was performed by comparing high temperature
areas to areas where first signs of wear appear, which further validates the tempera-
ture distribution. Bulk temperature measurements were used to validate the energy
partition between the cones as well as the cooling power in the test rig. However,
due to high thermal impedance in the interface between the thermocouples and the
bulk material, these measurements could not be used to validate the temperature
peak during synchronization. Measures were taken to increase the thermal conduc-
tance to the thermocouple, but the short synchronization time does not seem to
allow for accurate thermocouple measurement of peak temperatures.
6.2 Discussion
The author of this thesis has a background as a synchronizer design engineer, and
has often thought “If I, as a design engineer, was introduced to this thesis, how
would I use it?”’
• The pre-synchronization simulations in paper A highlight how dependent the

synchronizer is on the actuator and maneuvering system. Even though the
model is not validated, I would use the results of paper A to qualitatively
assess the effects from different parameters during design. In fact, the results
were successfully used as a guide when designing synchronizer grooves that
are now in production.
• The friction model presented in paper D is interesting for system simulations,
both for simulations with quick, simplified models as well as for multibody
dynamics simulations. This was successfully done by Muhammad Irfan in
his licentiate thesis [17]. Such simulations have previously proven valuable to
predict clash from external disturbances. However, at constant coefficient of
friction, some tweaking of the friction level was needed for clash to appear.
By using the modeled coefficient of friction, maybe the tweaks would not have
been necessary.
• Predicting failure from e.g. changed frictional properties due to adhesive wear
is, of course, highly valuable. If the surface temperature can be connected
to synchronizer failure, such models would significantly simplify synchronizer
development. The results from paper E have been used to predict the load
capacity of molybdenum coated synchronizers with a thinner molybdenum
coating. The results with thinner molybdenum coated were later validated
with physical testing in the µ-COMP rig. The loss of load capacity, i.e. at
what speed step the synchronizer failed, was predicted with high accuracy.
• Introducing a nominal relative cone angle looks promising, and could poten-
tially significantly increase the robustness and load capacity, especially for
6.3. FUTURE WORK 83
CFRP lined synchronizers. However, before introduction, an investigation at

higher load levels have to be performed. The computational cost for a full 3D
simulation of a synchronizer cycle is high, so an axi-symmetric model could
likely be used.
• Make more informed decisions regarding synchronizer tolerances, e.g. to not
increase the tolerance width of one cone just because it is possible to tighten
the tolerance on the other cone.
6.3 Future work

Some suggestions for future work are:
• More focus on long term effects, especially for CFRP lined synchronizers.
• The assessment of pre-synchronization function can be improved in some
ways. Improvements include an investigation of what the initial conditions
are, i.e. how much oil is present between the cones when the gear shift is
initiated, which likely varies with rotational speed. The effect of grooves
on pre-synchronization performance could be evaluated under realistic condi-
tions. Additionally, more validation would be interesting.
• Friction model for CFRP lined synchronizers with other types of transmission
oils.
• Measure the CFRP compressive modulus at different temperatures, and find
a more reliable way to estimate the temperature dependent coefficient of ther-
mal expansion.
References
[1] H. Ludanek, G. Sandgren, and F. Roos. “Drivetrain Functions for Reduc-

tion of CO2 -Emissions and higher Comfort in future Powertrain Concepts
with an integrated Approach”. In: VDI-Berichte 2256. Friedrichshafen: VDI
Wissensforum GmbH, 2015, pp. 3–20.
[2] G. Fontaras et al. Development of a CO2 certification and monitoring method-
ology for Heavy Duty Vehicles - Proof of Concept report. Tech. rep. Luxem-
bourg: European Commission Joint Research Centre, 2014.
[3] A. Leiserowitz et al. “International public opinion, perception, and under-
standing of global climate change”. In: Human development report (2007),
pp. 1–40.
[4] D. Robinette and D. Wehrwein. “Automatic Transmission Gear Ratio Op-
timization and Monte Carlo Simulation of Fuel Consumption with Parasitic
Loss Uncertainty”. In: SAE International Journal of Commercial Vehicles
8.2015-01-1145 (2015), pp. 45–62.
[5] R. Folkson. Alternative Fuels and Advanced Vehicle Technologies for Im-
proved Environmental Performance: Towards Zero Carbon Transportation.
Kidlington: Elsevier, 2014.
[6] S. Keidel et al. “Diesel engine fuel economy improvement enabled by super-
charging and downspeeding”. In: SAE International Journal of Commercial
Vehicles 5.2012-01-1941 (2012), pp. 483–493.
[7] T. Stoltz. “Powertrain Integration and Hybrids”. In: Workshop on Emerg-
ing Technologies for Heavy-Duty Fuel Efficiency. American Council for an
Energy-Efficient Economy and the International Council for Clean Trans-
portation. Washington, July 2014.
[8] D. Keski-Hynnila. “Engine Downsizing and Downspeeding”. In: Workshop
on Emerging Technologies for Heavy-Duty Fuel Efficiency. American Council
for an Energy-Efficient Economy and the International Council for Clean
Transportation. Washington, July 2014.
85
86 REFERENCES
[9] H Freudenthaler, O Hummer, and B Zhu. “Development of commercial ve-

hicle transmissions for Chinese market - potentials and risks”. In: VDI-
Berichte 2187. Friedrichshafen: VDI Wissensforum GmbH, 2013, pp. 499–
502.
[10] Volvo Group Trucks. Volvo Trucks launches a unique gearbox for heavy ve-
hicles. Press release. Göteborg. 2014, Jul. 22.
[11] O. Meyer-Rühle et al. Statistical coverage and economic analysis of the lo-
gistics sector in the EU (SEALS). Tech. rep. Prepared for the European
Commission, DG Energy and Transport, 2008.
[12] A. Moawad and A. Rousseau. Impact of Transmission Technologies on Fuel
Efficiency to Support 2017-2025 CAFE Regulations. Tech. rep. SAE Tech-
nical Paper 2014-01-1082, 2014.
[13] R. Acuner. “Synchronisierungen mit Carbon-Reibwerkstoffen unter hohen
und extremen Beanspruchungen”. PhD thesis. Technische Universität München,
Oct. 2016.
[14] S. Neudörfer. “Thermomechanische Einflüsse auf die Tribologie von Synchro-
nisierungen”. PhD thesis. Gottfried Wilhelm Leibniz Universität, Hannover,
Aug. 2008.
[15] T. Lösche, E.-G. Paland, and G. Poll. “Wear Behaviour of Synchronisers
in Relation to a Duty Parameter”. In: Tribology for Energy Conservation.
Vol. 34. Tribology Series. Elsevier, 1998, pp. 95 –102.
[16] T. Kinugasa et al. “Thermal analysis of the synchronizer friction surface
and its application to the synchronizer durability improvement”. In: JSAE
review 20.2 (1999), pp. 217–222.
[17] M. Irfan. “Modelling and optimization of gear shifting mechanism, Appli-
cation to heavy vehicles transmission systems”. Licentiate Thesis. Chalmers
University of Technology, Göteborg, Jan. 2017.
[18] Scania CV AB. Scania Annual Report 2014. Södertälje. 2015, Mar. 26.
[19] M. Andersson. “An experimental investigation of spur gear efficiency and
temperature: A comparison between ground and superfinished surfaces”.
PhD thesis. KTH Royal Institute of Technology, Stockholm, Mar. 2017.
[20] E. Hartono. “Churning losses and efficiency in gearboxes”. Licentiate Thesis.
Chalmers University of Technology, Göteborg, Nov. 2014.
[21] M. Sosa. “Running-in of gears-surface and efficiency transformation”. PhD
thesis. KTH Royal Institute of Technology, Stockholm, Nov. 2017.
[22] D. Mallipeddi. “Surface Integrity Characterization of Gears with respect to
Running-in”. Licentiate Thesis. Chalmers University of Technology, Göte-
borg, Aug. 2016.
[23] L. Wramner. “Torsional vibrations in truck powertrains with dual mass fly-
wheel having piecewise linear stiffness”. In: ENOC 2017. Budapest, 2017.
REFERENCES 87
[24] E. Gomez. “Multiple order excitation and response of centrifugal pendulum

vibration absorbers”. In: To appear in ISMA 2018 - International Conference
on Noise and Vibration Engineering. Leuven, 2018.
[25] European Commission. Directive 70/156/EEC, Annex 2. Accessed 2015,
Apr. 17. 1970, Feb. 23.
[26] United States Environmental Protection Agency. Vehicle Weight Classifica-
tions | Emission Standards Reference Guide. Accessed 2015, Apr. 27. 2012,
Nov. 14.
[27] S. H. Thomke. “Simulation, learning and R&D performance: Evidence from
automotive development”. In: Research Policy 27.1 (1998), pp. 55–74.
[28] U. Sellgren. “Simulation driven design-Necessity and feasibility”. Depart-
ment of Machine Design, Royal Institute of Technology, Sweden. 1994.
[29] D. M. Baskin et al. A case study in structural optimization of an automotive
body-in-white design. Tech. rep. SAE Technical Paper 2008-01-0880, 2008.
[30] J. Will. “State of the Art–robustness evaluation in CAE-based virtual proto-
typing processes of automotive applications”. In: Proceedings Optimization
and Stochastic Days 4 (2007).
[31] J. Wang et al. “Topology Optimization of Gearbox to Reduce Radiated
Noise”. In: ASME 2015 International Design Engineering Technical Confer-
ences and Computers and Information in Engineering Conference. American
Society of Mechanical Engineers. 2015.
[32] F. E. Kennedy. “Thermal and thermomechanical effects in dry sliding”. In:
Wear 100 (1984), pp. 453–476.
[33] U. Sellgren. “Simulation-driven design: motives, means, and opportunities”.
PhD thesis. KTH Royal Institute of Technology, Stockholm, Dec. 1999.
[34] W. L. Oberkampf and C. J. Roy. Verification and validation in scientific
computing. New York: Cambridge University Press, 2010.
[35] W. L. Oberkampf and T. G. Trucano. “Verification and validation bench-
marks”. In: Nuclear engineering and Design 238.3 (2008), pp. 716–743.
[36] J. P. Kleijnen. “Verification and validation of simulation models”. In: Euro-
pean journal of operational research 82.1 (1995), pp. 145–162.
[37] E. Tadmor. “A review of numerical methods for nonlinear partial differential
equations”. In: Bulletin of the American Mathematical Society 49.4 (2012),
pp. 507–554.
[38] N. S. Ottosen and H. Petersson. Introduction to the finite element method.
Harlow: Pearson Education Limited, 1992.
[39] Y.-B. Yang and M.-S. Shieh. “Solution method for nonlinear problems with
multiple critical points”. In: AIAA journal 28.12 (1990), pp. 2110–2116.
88 REFERENCES
[40] J. C. Butcher. Numerical methods for ordinary differential equations. Corn-

wall: John Wiley & Sons, 2008.
[41] T. J. Hughes, J. A. Cottrell, and Y. Bazilevs. “Isogeometric analysis: CAD,
finite elements, NURBS, exact geometry and mesh refinement”. In: Com-
puter methods in applied mechanics and engineering 194.39 (2005), pp. 4135–
4195.
[42] L. Phillips. Cars 1895-1965. Bloomington: Xlibris Corporation, 2011.
[43] Volvo Trucks. Feature: Technology comes full circle - history of the gearbox.
Press information. Göteborg. 2010, Nov. 1.
[44] M. A. Wirth. “Schleppmomente in Synchronisierungen von Fahrzeuggetrieben”.
PhD thesis. Technische Universität München, Nov. 2012.
[45] K Stahl et al. “Function of Synchronizer Systems at Low Temperatures”. In:
15th International Conference on Experimental Mechanics. Porto, 2012.
[46] L. M. Sykes. The Jaguar XJ220 Triple-Cone Synchronizer A Case Study.
Tech. rep. SAE Technical Paper 940737, 1994.
[47] D. Häggström and M. Nordlander. “Development of a Program for Calcu-
lating Gearbox Synchronization”. Master thesis. Luleå University of Tech-
nology. Luleå, Sweden, Feb. 2011.
[48] L Lovas et al. “Mechanical behaviour simulation for synchromesh mechanism
improvements”. In: Proceedings of the Institution of Mechanical Engineers,
Part D: Journal of Automobile Engineering 220.7 (2006), pp. 919–945.
[49] B Paffoni et al. “The hydrodynamic phase of gearbox synchromesh oper-
ation: the influence of radial and circumferential grooves”. In: Proceedings
of the Institution of Mechanical Engineers, Part J: Journal of Engineering
Tribology 211.2 (1997), pp. 107–116.
[50] G. Moir. “An investigation into objective measures of gear-shift quality”. In:
Proceedings of the Institution of Mechanical Engineers, Part D: Journal of
Automobile Engineering 209.4 (1995), pp. 273–279.
[51] M. Spreckels. “Einfluss der Temperaturverteilung auf das tribologische Ver-
halten von Synchronisierungen”. PhD thesis. Gottfried Wilhelm Leibniz Uni-
versität, Hannover, Dec. 2001.
[52] G. Poll and M. Spreckels. “Influence of temperature distribution on the tribo-
logical performance of automotive synchronisers”. In: Tribological Research
and Design for Engineering Systems. Vol. 41. Tribology Series. Elsevier,
2003, pp. 613 –621.
[53] R Acuner et al. “Influence of Cone-Angle-Difference on Performance of Syn-
chronizers with Carbon Friction Linings”. In: VDI-Berichte 2218. Friedrichshafen:
VDI Wissensforum GmbH, 2014, pp. 595–610.
[54] FVA GmbH. About us. Accessed 2017-10-18. url: https://www.fva-serv
ice.de/en/about/about-us/.
REFERENCES 89
[55] A. Bouffet. Evaluating tribology of synchronizers for today’s manual trans-

missions. Tech. rep. SAE Technical Paper 2004-01-2022, 2004.
[56] G. Brown et al. “Understanding MTF Additive Effects on Synchroniser Fric-
tion”. In: SAE International Journal of Fuels and Lubricants 5.1 (2011),
pp. 447–458.
[57] G. Walker et al. Understanding MTF Additive Effects on Synchroniser Fric-
tion - Part 2, Structure Performance Analysis. Tech. rep. SAE Technical
paper2012-01-1668, 2012.
[58] O. Gangvekar and S. Deshmane. Grit Blasting on Synchronizer-To Resolve
Early Crashing Complaint. Tech. rep. SAE Technical Paper 2017-01-1769,
2017.
[59] P. D. Walker. “Synchroniser analysis and shift dynamics of powertrains
equipped with dual clutch transmissions”. PhD thesis. University of Tech-
nology, Sydney, July 2011.
[60] H. V. Alizadeh, M. K. Helwa, and B. Boulet. “Modeling, analysis and con-
strained control of wet cone clutch systems: A synchromesh case study”. In:
Mechatronics 49 (2018), pp. 92 –104.
[61] E. M’Ewen. “The theory of gear-changing”. In: Proceedings of the Institution
of Mechanical Engineers: Automobile Division 3.1 (1949), pp. 30–40.
[62] The Coordinating European Council. Evaluation of the Synchromesh En-
durance Life using the FZG SSP 180 synchromesh test rig (L-66-99 Issue
5.1). Brussels. 2014, Aug. 1.
[63] T. Saito et al. Study of Durability Prediction with Focus on Wear Properties
for Multiple Plate Clutches. Tech. rep. SAE Technical paper 2007-01-0240,
2007.
[64] S. Ohkawa et al. Elasticity - an important factor of wet friction materials.
Tech. rep. SAE Technical Paper 911775, 1991.
[65] K. Okabe, A. Fujimoto, and T. Harasawa. Proposal of Field Life Design
Method for Wet Multiple Plate Clutches of Automatic Transmission on Forklift-
trucks. Tech. rep. SAE Technical paper 2009-01-2934, 2009.
[66] T. Hirano et al. Development of Friction Material and Quantitative Analysis
for Hot Spot Phenomenon in Wet Clutch System. Tech. rep. SAE Technical
paper 2007-01-0242, 2007.
[67] P. Marklund et al. “Thermal influence on torque transfer of wet clutches in
limited slip differential applications”. In: Tribology international 40.5 (2007),
pp. 876–884.
[68] P. Marklund and R. Larsson. “Wet clutch friction characteristics obtained
from simplified pin on disc test”. In: Tribology International 41.9 (2008),
pp. 824–830.
90 REFERENCES
[69] T. Saito et al. Technology for Prediction of Torque Capacity during Operation
of AT with Focus on Friction Properties of Facing Materials in Multiple Plate
Clutches. Tech. rep. SAE Technical paper 2009-01-1255, 2009.
[70] R Rank and A Kearsey. “Carbon Based friction materials for automotive
applications”. In: 14th International Colloquium Tribology. Esslingen, 2004.
[71] R. C. Oldfield and R. F. Watts. “Impact of lubricant formulation on the
friction properties of carbon fiber clutch plates”. In: Lubrication Science
18.1 (2006), pp. 37–48.
[72] Ricardo UK Ltd. Ricardo Gearshift Quality Assessment Manual. 2007.
[73] J. Kim et al. Development of shift feeling simulator for a manual transmis-
sion. Tech. rep. SAE Technical Paper 2002-01-2202, 2002.
[74] H. Hoshino. “Simulation on synchronization mechanism of transmission gear-
box”. In: International ADAMS User Conference. 1998.
[75] Y.-C. Liu and C.-H. Tseng. “Simulation and analysis of synchronisation and
engagement on manual transmission gearbox”. In: International journal of
vehicle design 43.1-4 (2007), pp. 200–220.
[76] R. Socin and K. Walters. Manual transmission synchronizers. Tech. rep.
SAE Technical Paper 680008, 1968.
[77] N. Abdel-Halim et al. “Performance of multicone synchronizers for manual
transmissions”. In: Proceedings of the Institution of Mechanical Engineers,
Part D: Journal of Automobile Engineering 214.1 (2000), pp. 55–65.
[78] P. D. Walker and N. Zhang. “Parameter study of synchroniser mechanisms
applied to dual clutch transmissions”. In: International Journal of Power-
trains 1.2 (2011), pp. 198–220.
[79] O. Back. Basics of Synchronizers. Tech. rep. Hoerbiger Antriebstechnik
GmbH, 2013.
[80] S. Technologies. Detents ARRES Presynchronisation of transmissions. TPI
178.
[81] Hoerbiger Antriebstechnik GmbH. Hoerbiger Friction Systems - Ideal solu-
tions for any application. Accessed 2018-01-23. url: http://www.hoerbige
r.com/upload/file/hoerbigerfrictionsystems-idealsolutionsforan
yapplication.pdf.
[82] Schaeffler Technologies GmbH. “Holistic Development of Synchronizing Sys-
tems”. In: Schaeffler Technologies GmbH & Co. KG (eds) Solving the Pow-
ertrain Puzzle. Wiesbaden: Springer Vieweg, 2014.
[83] SGL Group. Sigracomp WF - Wet friction material for transmission syn-
chronizers. Accessed 2016-03-17. url: http://www.sglgroup.com/cms/
_common / downloads / products / product - groups / cc / wet - friction -
materials/SIGRACOMP_WF_e.pdf.
REFERENCES 91
[84] Oerlikon Metco. Synchronizer Components. Accessed 2018-01-23. url: ht

tps://www.oerlikon.com/metco/en/products- services/friction-
systems/synchronizer-components/.
[85] W Augustin. “Friction elements running in oil”. Pat. US3927241. 1974.
[86] J. Fei et al. “Carbon-fiber reinforced paper-based friction material: study on
friction stability as a function of operating variables.(Author abstract)(Report)”.
In: Journal of Tribology 130.4 (2008).
[87] M. Li et al. “On the wear prediction of the paper-based friction materialin
a wet clutch”. In: Wear 334-335.C (2015), pp. 56–66.
[88] A Gardziella, L. A. Pilato, and K. A. Phenolic resins : chemistry, applica-
tions, standardization, safety and ecology. 2nd completely revised edition.
Berlin: Springer, 2000.
[89] B. T. Åström. Manufacturing of polymer composites. Cheltenham: Nelson
Thornes Ltd, 1997.
[90] P. S. Winckler. “Carbon-based friction material for automotive continuous
slip service”. Pat. US5662993. 1997.
[91] X. Huang. “Fabrication and Properties of Carbon Fibers”. In: Materials 2.4
(2009), pp. 2369–2403.
[92] D Edie. “The effect of processing on the structure and properties of carbon
fibers”. In: Carbon (UK) 36.4 (1998), pp. 345–362.
[93] Geometrical product specifications (GPS) – Geometrical tolerancing – Tol-
erances of form, orientation, location and run-out. Standard. Geneva, CH:
International Organization for Standardization, Feb. 2017.
[94] G. Bóka et al. “Face dog clutch engagement at low mismatch speed”. In:
Periodica Polytechnica. Transportation Engineering 38.1 (2010), pp. 29–35.
[95] SK Hydroautomation GmbH. Test rigs. Accessed 2014, Apr. 9. url: https:
//www.sk-hydroautomation.de/index.php?id=pruefstaende&lang=en.
[96] T. G. Ore, R. A. Nellums, and G. Skotnicki. Improved synchronizers for
truck transmissions. Tech. rep. SAE Technical Paper 952602, 1995.
[97] G. Bóka. “Shifting optimization of face dog clutches in heavy duty auto-
mated mechanical”. PhD thesis. Budapest University of Technology and
Economics, Nov. 2011.
[98] H. Liu et al. “Gear-shift strategy for a clutchless automated manual trans-
mission in battery electric vehicles”. In: SAE International Journal of Com-
mercial Vehicles 5.2012-01-0115 (2012), pp. 57–62.
[99] Scania CV AB. Freewheeling retarder reduces fuel consumption. Press re-
lease. Södertälje. 2014, May 14.
92 REFERENCES
[100] E. Bartos and K. Ahlberg. “Minimizing of Drain Leakage on a Scania Re-

tarder”. Master thesis. KTH Royal Institute of Technology.Stockholm, Swe-
den, May 2011.
[101] A. van Beek. Advanced engineering design. Delft: TU Delft, 2009.
[102] B O’Connor et al. Influence of Additive Chemistry on Manual Transmission
Synchronizer Performance. Tech. rep. SAE Technical Paper 2002-01-1697,
2002.
[103] R. Stribeck. “Die wesentliche Eigenschaften der Gleit-und Rollenlager”. In:
VDI-Zeitschrift 36 (1902), pp. 1341–1348.
[104] X. Lu, M. Khonsari, and E. Gelinck. “The Stribeck curve: experimental
results and theoretical prediction”. In: Journal of tribology 128.4 (2006),
pp. 789–794.
[105] T. R. Thomas. Rough surfaces. London: Imperial College Press, 1999.
[106] J. Greenwood and J. Williamson. “Contact of nominally flat surfaces”. In:
Proceedings of the Royal Society of London A: Mathematical, Physical and
Engineering Sciences. Vol. 295. 1442. The Royal Society. 1966, pp. 300–319.
[107] K. L. Johnson. Contact mechanics. Cambridge: Cambridge University Press,
1985.
[108] E. Bergseth, S. Sjöberg, and S. Björklund. “Influence of real surface topog-
raphy on the contact area ratio in differently manufactured spur gears”. In:
Tribology International 56 (2012), pp. 72–80.
[109] H Uetz and J Föhl. “Wear as an energy transformation process”. In: Wear
49 (1978), pp. 253–264.
[110] M. Amiri and M. M. Khonsari. “On the thermodynamics of friction and
wear - a review”. In: Entropy 12.5 (2010), pp. 1021–1049.
[111] F. E. Kennedy. “Single pass rub phenomena - analysis and experiment”. In:
Journal of Lubrication Technology 104 (1982), pp. 582–588.
[112] H. Blok. “The flash temperature concept”. In: Wear 6.6 (1963), pp. 483 –
494.
[113] A. Anderson and R. Knapp. “Hot spotting in automotive friction systems”.
In: Wear 135.2 (1990), pp. 319–337.
[114] G. Sutter and N. Ranc. “Flash temperature measurement during dry friction
process at high sliding speed”. In: Wear 268.11 (2010), pp. 1237 –1242.
[115] B Bhushan. “Magnetic Head-Media Interface Temperatures: Part 1—Anal-
ysis”. In: Journal of tribology 109.2 (2002), pp. 243–251.
[116] X. Tian and F. E. Kennedy. “Contact surface temperature models for finite
bodies in dry and boundary lubricated sliding”. In: Journal of tribology 115.3
(1993), pp. 411–418.
REFERENCES 93
[117] A Soom, C. Serpe, and G. Dargush. “Thermomechanics of sliding contact”.

In: Proceedings of the NATO Advanced Study Institute on Fundamentals of
Tribology and Bridging the Gap Between the Macro- and Micro/Nanoscales.
Keszthely, 2000, pp. 467–485.
[118] J. Barber. “The influence of thermal expansion on the friction and wear
process”. In: Wear 10.2 (1967), pp. 155 –159.
[119] R. Burton. “Thermal deformation in frictionally heated contact”. In: Wear
59 (1980), pp. 1–20.
[120] J. R. Barber. “Thermoelastic instabilities in the sliding of conforming solids”.
In: Proceedings of the Royal Society of London. Series A, Mathematical and
Physical Sciences 312.1510 (1969), pp. 381–394.
[121] S. Panier, P. Dufrénoy, and D. Weichert. “An experimental investigation of
hot spots in railway disc brakes”. In: Wear 256.7 (2004), pp. 764–773.
[122] J. Barber. “Thermoelastic instabilities in the sliding of comforming solids”.
In: Proceedings of the Royal Society of London A: Mathematical, Physical
and Engineering Sciences. Vol. 312. 1510. The Royal Society. 1969, pp. 381–
394.
[123] T. A. Dow and R. Burton. “Thermoelastic instability of sliding contact in
the absence of wear”. In: Wear 19.3 (1972), pp. 315–328.
[124] K. Lee and J. Barber. “Frictionally excited thermoelastic instability in au-
tomotive disk brakes”. In: Journal of tribology 115.4 (1993), pp. 607–614.
[125] P. Zagrodzki and S. A. Truncone. “Generation of hot spots in a wet multidisk
clutch during short-term engagement”. In: Wear 254.5 (2003), pp. 474–491.
[126] J. Jang and M. Khonsari. “On the formation of hot spots in wet clutch
systems”. In: Journal of tribology 124.2 (2002), pp. 336–345.
[127] Dassault Systèmes Simulia Corp. Abaqus [2016] Theory Guide - 5.2.1 Con-
tact pressure definition.
[128] A Almqvist and F Perez Rafols. “An LCP based approach for the contact
mechanics of elastic half spaces”. In: XXIV ICTAM. Montreal, 2016.
[129] Y. Kanno, J. Martins, and A Pinto da Costa. “Three-dimensional quasi-
static frictional contact by using second-order cone linear complementarity
problem”. In: International journal for numerical methods in engineering
65.1 (2006), pp. 62–83.
[130] F Tin-Loi and S. Xia. “An iterative complementarity approach for elasto-
plastic analysis involving frictional contact”. In: International Journal of
Mechanical Sciences 45.2 (2003), pp. 197–216.
[131] Dassault Systèmes Simulia Corp. Abaqus [2016] Analysis User’s Guide -
38.1.3 Smoothing contact surfaces in Abaqus/Standard.
94 REFERENCES
[132] Dassault Systèmes Simulia Corp. Best Practices for Contact Modeling for
Accuracy and Accelerated Convergence. Course held by Dassault Systèmes
Simulia Corp.
38.1.1 Contact formulations in Abaqus/Standard.
38.1.2 Contact constraint enforcement methods in Abaqus/Standard.
[135] N.-H. Kim. Introduction to nonlinear finite element analysis. New York:
Springer Science & Business Media, 2014.
[136] Dassault Systèmes Simulia Corp. Abaqus [2016] Theory Guide - 2.2.1 Non-
linear solution methods in Abaqus/Standard.
[137] COMSOL AB. Solving Nonlinear Static Finite Element Problems. Accessed
2018-02-02. url: https://www.comsol.com/blogs/solving-nonlinear-
static-finite-element-problems/.
[138] M. Söderfjäll. “Friction in Piston Ring-Cylinder Liner Contacts”. PhD thesis.
Luleå University of Technology, Apr. 2017.
[139] M. Mileti, H. Pflaum, and K. Stahl. “TorqueLINE Cone Clutch: Thermo-
Mechanical Stability of Cone Clutches for ATs”. In: VDI-Berichte 2328.
Friedrichshafen: VDI Wissensforum GmbH, 2018, pp. 357–368.
[140] J Schnittger and A Folkesson. Funktionsanalys och optimering i maskin-
tekniken (in Swedish). Stockholm: KTH Royal Institute of Technology, 1998.
[141] Plastics – Differential scanning calorimetry (DSC) – Part 1: General princi-
ples. Standard. Geneva, CH: International Organization for Standardization,
Sept. 2016.
[142] J. A. Cape and G. W. Lehman. “Temperature and Finite Pulse-Time Effects
in the Flash Method for Measuring Thermal Diffusivity”. In: Journal of
Applied Physics 34.7 (1963), pp. 1909–1913.
[143] C. Pradere et al. “Thermal properties of carbon fibers at very high temper-
ature”. In: Carbon 47.3 (2009), pp. 737–743.
[144] V. A. Erä, A. Mattila, and J. J. Lindberg. “Determination of specific heat
of phenol formaldehyde resol resins by differential scanning calorimetry”. In:
Angewandte Makromolekulare Chemie 64.1 (1977), pp. 235–238.

FULLTEXT01

Uploaded by

Copyright:

Available Formats

FULLTEXT01

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

FULLTEXT01

Uploaded by

Copyright:

Available Formats

Towards model-based development of heavy truck

Keywords: synchronization, synchronizers, gear shift, gearbox

Växellådans växlingsprestanda får allt större betydelse. Synkroniserings-

Nyckelord: synkronisering, synkroniseringsenheter, växling, växellåda

3.2.4 Integrating simulations and physical tests . . . . . . . . . . . 48

5 Summary of appended papers 75

6 Conclusions, discussion and future work 79

This thesis consists of a summary and the following seven papers:

Publications not included in this thesis

AMT Automated manual transmission

Alternative part naming

A Out-of-roundness amplitude [m]

U Sliding speed, simplified 3D FSI model [m/s]

1.1 Research objective and motivation

∗ Throttle and brake pedal, no clutch pedal

1.2 Research questions

The main research question is divided into the following sub-questions:

• RQ1: How can the frictional behavior at transition between pre-synchronization

• RQ3: How do external loads as well as different characteristic design pa-

• RQ6: Is there a surface temperature in a synchronizer with CFRP lining,

• RQ7: How do manufacturing tolerances affect the synchronizers maximum

• RQ8: How can thermomechanical synchronization models be validated?

1.3 Thesis outline

Virtual development of Virtual assessment of

Figure 1.1: Contributions to the overall goal by the appended papers.

1.4 Swedish transmission cluster

This chapter provides background information to the presented research. Section

2.1 Simulation-driven product development

2.2 Gear shift in heavy truck transmissions

Main gearbox Low range High range

2.3 Synchronizer background information

(a) (b) (c)

2.3.1 Related work

∗ Forschungsvereinigung Antriebstechnik, the (German) Research Association for Drive Tech-

synchronization torque must increase by n3 to achieve similar synchronization time.

Unit Truck Passenger car Ratio

2.3.2 Different phases of synchronization

1. Neutral. Initially, the shift sleeve is in the neutral position (around 50 %

2. Pre-synchronization. The shift sleeve leaves the neutral position to ener-

3. Main synchronization. The rotational speed is synchronized. During this

4. Blocker release. Synchronous rotational speed is reached, and the latch

5. Gear engagement. The shift sleeve is moved to an engaged position. Since

(b) (c) (d) (e) (f)

Many synchronization process definitions also include gear disengagement before

2.3.3 Preventing unsynchronized gear engagement

(a) Rotational clearance (b)

2.3.4 Pre-synchronization phase

Region where spring system Region where blocking is

The transition between pre-synchronization phase and main synchronization

2.3.5 Main synchronization phase

• Rotational speed difference to synchronize [14, 51].

• Moment of inertia to synchronize. The moment of inertia of every component

IR is the reflected moment of inertia, Iz is the moment of inertia of component

• Drag torque, which either assists or counteracts synchronization depending

Effect of vehicle acceleration on synchronizer

500 Gear wheel

• Specific energy to synchronize, i.e. energy per (nominal) unit area.

These duty parameters are defined in paper E.

2.3.6 Synchronizer wear and wear gap