Modeling and Control of an Automated Manual Transmission for EcoCAR 3 Vehicle

THESIS

Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in
the Graduate School of The Ohio State University

By

Aditya Modak

Graduate Program in Mechanical Engineering

The Ohio State University

2017

Master's Examination Committee:

Dr. Shawn Midlam-Mohler (Advisor)

Dr. Giorgio Rizzoni

Copyrighted by

Aditya Modak

2017
 Abstract

EcoCAR 3 is a part of the Advanced Vehicle Technology Competition series hosted

by the Department of Energy, and it challenges 16 North American university teams to re-

engineer a 2016 Chevrolet Camaro and turn it into a hybrid electric vehicle, thus improving

the environmental impact of the car while retaining its performance aspects.

The Ohio State University’s EcoCAR 3 vehicle has a plug-in hybrid architecture,

with operation in series and parallel power flows. The architecture features a 5-speed

manual transmission that was automated by the team to retain the efficiency of a manual

transmission while providing the convenience of an automatic transmission. The team-

developed controllers manage the clutch and shift actuators to provide supervisory control

of the automated manual transmission.

The simplicity and efficiency of a manual transmission combined with the

advantages provided by the hybrid architecture make it a good candidate for an HEV. This

thesis provides an overview of the modeling, component testing, and controls development

for the AMT system. The controls development includes high level control for vehicle

launch, gearshift process, and strategies used in different hybrid vehicle operation modes.

ii
To my family for their ceaseless support and encouragement throughout my education.

ज्ञानं परमं बलम ्

iii
 Acknowledgments

This thesis is meaningless if I do not thank the entities responsible for making it

happen. OSU’s Center for Automotive Research hosts many talented people who helped

with the work done in this thesis. I thank Prof. Shawn Midlam-Mohler, a phenomenal

advisor, for his guidance and immensely practical advice over the last two years, and Prof.

Giorgio Rizzoni, for his support and insights during my tenure at CAR.

I would like to acknowledge Arjun Khanna for mentoring me when I started out on

the team, and Gregory Jankord for being a fantastic vehicle testing partner. I thank Simon

Trask and Brandon Bishop for doing the spadework to set up the test rigs and dealing with

transmission fluid, and Guido Guercioni for his advice and insights. I am thankful to

Dennis Kibalama and Andrew Johnson for the endless time they spent doing the wiring

and electrical troubleshooting, and for keeping a straight face while watching me work on

electrical things. A big shout-out to the team leads and members for their efforts: Andrew

Huster, Wilson Perez, Philip Dalke, Andrew Fu, Ullekh Gambhira, Kepin Kavathia, Aditya

Karumanchi, Zhaoxuan Zhu, Kristina Kuwabara, Vinay Ravi, Samuel Yacinthe, M.J.

Yatkso, Nick Tomczak and Benedict Sim.

Finally, I am thankful to the competition and the sponsors for giving many others

like me the opportunity to work with advanced automotive powertrains.

iv
 Vita

January 1, 1992 ..............................................Born – Indore, India

2009................................................................D.B.J. College, Chiplun, India

2013................................................................B.E. (Hons.) Mechanical Engineering,

BITS Pilani Goa, India

2015 to present ..............................................Graduate Research Associate,

The Ohio State University

Publications

Kibalama, D., Huster, A., Khanna, A., Modak, A. et al., "Testing and Validation of a Belted
Alternator System for a Post-Transmission Parallel PHEV for the EcoCAR 3 Competition,"
SAE Technical Paper 2017-01-1263, 2017, doi:10.4271/2017-01-1263.

Trask, S., Jankord, G., Modak, A., et al., “System Diagnosis and Fault Mitigation Strategy
Development for an Automated Manual Transmission using Structural Analysis,” ASME
Dynamic Systems and Controls Conference, Tysons Corner, 2017.

Fields of Study

Major Field: Mechanical Engineering

v
 Table of Contents

Abstract ............................................................................................................................... ii

Acknowledgments.............................................................................................................. iv

Vita...................................................................................................................................... v

Publications ......................................................................................................................... v

Fields of Study .................................................................................................................... v

Table of Contents ............................................................................................................... vi

List of Tables ...................................................................................................................... x

List of Figures .................................................................................................................... xi

Chapter 1: Introduction ....................................................................................................... 1

1.1 Background .......................................................................................................... 1

1.2 EcoCAR 3 ............................................................................................................ 2

1.3 Thesis Objective ................................................................................................... 5

1.4 Thesis Organization.............................................................................................. 5

Chapter 2: Literature Review .............................................................................................. 6

2.1 Introduction .......................................................................................................... 6

vi
 2.2 Transmissions Overview ...................................................................................... 6

2.2.1 Manual Transmission .................................................................................... 6

2.2.2 Automatic Transmission ............................................................................... 8

2.2.3 Dual Clutch Transmission............................................................................. 9

2.2.4 Continuously Variable Transmission ............................................................ 9

2.2.5 Automated Manual Transmission ............................................................... 10

2.3 AMT Control ...................................................................................................... 11

2.3.1 Vehicle Launch ........................................................................................... 12

2.3.2 Gearshift Sequence ..................................................................................... 14

2.3.3 Shift Scheduling and Drive Quality ............................................................ 16

Chapter 3: Component Description .................................................................................. 18

3.1 Introduction ........................................................................................................ 18

3.2 Clutching Subsystem .......................................................................................... 20

3.3 Shifting Subsystem ............................................................................................. 21

Chapter 4: Modeling ......................................................................................................... 26

4.1 Introduction ........................................................................................................ 26

4.2 Modeling Approach............................................................................................ 28

4.3 Clutching Subsystem .......................................................................................... 31

4.4 Shifting Subsystem ............................................................................................. 36

vii
Chapter 5: Component Level Testing ............................................................................... 45

5.1 Introduction ........................................................................................................ 45

5.2 Controller I/O Testing ........................................................................................ 45

5.3 Shifting Testing .................................................................................................. 47

5.4 Clutch Testing .................................................................................................... 51

Chapter 6: Control............................................................................................................. 56

6.1 Introduction ........................................................................................................ 56

6.2 Shifting Logic Overview .................................................................................... 56

6.3 GCM Shifting Logic........................................................................................... 59

6.4 Clutch Initialization Procedure........................................................................... 62

6.5 Bite Point Finding Procedure ............................................................................. 65

6.6 Clutch Engagement Strategies ........................................................................... 66

6.6.1 Clutch Engagement Types .......................................................................... 67

6.6.2 Vehicle Launch ........................................................................................... 68

6.6.3 Performance Launch ................................................................................... 78

6.7 Vehicle Mode Based AMT management ........................................................... 81

6.7.1 Vehicle Startup............................................................................................ 82

6.7.2 EV Only Mode ............................................................................................ 82

6.7.3 Charge Sustaining Mode ............................................................................. 82

viii
 6.7.4 Performance Mode ...................................................................................... 84

6.7.5 Series Mode ................................................................................................ 87

6.7.6 Engine Only Mode ...................................................................................... 88

6.7.7 Vehicle Shutdown ....................................................................................... 90

6.7.8 Parked Charging Mode ............................................................................... 90

6.7.9 Miscellaneous Modes.................................................................................. 90

Chapter 7: Conclusions and Future Work ......................................................................... 91

7.1 Accomplishments and Challenges ..................................................................... 91

7.2 Future Work ....................................................................................................... 92

Bibliography ..................................................................................................................... 93

ix
 List of Tables

Table 1: OSU EcoCAR 3 vehicle powertrain component specifications ........................... 4

Table 2: Typical gear efficiencies for a manual transmission [2] ....................................... 7

Table 3: Typical gear efficiencies for an automatic transmission [2] ................................. 8

Table 4: CCU softECU I/O selection................................................................................ 31

Table 5: Clutching subsystem components....................................................................... 32

Table 6: CCU operation states .......................................................................................... 33

Table 7: CCU I/O signals .................................................................................................. 33

Table 8: SCU I/O signals .................................................................................................. 37

Table 9: SCU operating states........................................................................................... 38

Table 10: Shift linkage positions ...................................................................................... 41

Table 11: Comparison of CIP estimated disengagement positions .................................. 64

Table 12: Important signals used by counter based launch method ................................. 70

Table 13: Comparison of gear ratios for OSU EcoCAR and stock Camaro ..................... 73

x
 List of Figures

Figure 1: EcoCAR 3 vehicle development process ............................................................ 2

Figure 2: OSU EcoCAR 3 vehicle architecture .................................................................. 3

Figure 3: Internal structure of a manual transmission [3] ................................................... 7

Figure 4: Schematic of DCT, ZF 6 speed AT [4] ............................................................... 9

Figure 5: Classification of AMTs [6]................................................................................ 11

Figure 6: Typical clutch engagement trajectory, actual launch at 30% APP [8] .............. 13

Figure 7: Diaphragm spring force vs displacement [10], bearing load vs travel [11] ...... 14

Figure 8: Gearshift positions and control phases [15] ...................................................... 15

Figure 9: Components in the AMT system ....................................................................... 18

Figure 10: Location of AMT system components in OSU EcoCAR 3 vehicle ................ 19

Figure 11: Clutch actuation system................................................................................... 21

Figure 12: SCU actuator lever arrangement ..................................................................... 22

Figure 13: Actuation sequence for 2-3 upshift ................................................................. 23

Figure 14: Modular structure of EcoSIM 3....................................................................... 27

Figure 15: Component level structure of EcoSIM 3 ......................................................... 28

Figure 16: V diagram for MBD process ........................................................................... 28

Figure 17: Component model development summary ...................................................... 30

Figure 18: CCU operation states ....................................................................................... 33

xi
Figure 19: Motor position – line pressure behaviour with varying air amounts ............... 35

Figure 20: Clutch subsystem model summary .................................................................. 36

Figure 21: Addition of pull-up resistors for HIL simulator wiring harness ...................... 38

Figure 22: Transmission model with double-sided synchronizers ................................... 40

Figure 23: Gear shift actuating mechanism ...................................................................... 41

Figure 24: Shift linkage positions for 2-3 upshift ............................................................. 42

Figure 25: Shift time comparison for 3-4 upshift ............................................................. 43

Figure 26: Shifting subsystem model summary................................................................ 44

Figure 27: SCU I/O testing setup ...................................................................................... 47

Figure 28: Potentiometer reported values on shifting test rig ........................................... 50

Figure 29: Torque transmissibility test rig ........................................................................ 52

Figure 30: Motor position – line pressure plots on different setups ................................. 53

Figure 31: Position – torque and pressure – torque characteristics .................................. 54

Figure 32: Clutch transmissible torque plotted against percent engagement ................... 55

Figure 33: Overall shifting logic in HSC .......................................................................... 57

Figure 34: Shifting process events for a 3-4 upshift in engine only mode ....................... 59

Figure 35: Shifting logic in GCM ..................................................................................... 60

Figure 36: Blip generation by GCM ................................................................................. 61

Figure 37: CCU startup and shutdown handshakes .......................................................... 62

Figure 38: Execution of BFP on vehicle ........................................................................... 66

Figure 39: Counter based launch algorithm structure ....................................................... 69

Figure 40: Clutch speed based release counter rate .......................................................... 71

xii
Figure 41: Vehicle launch with engine only ..................................................................... 72

Figure 42: Clutch pedal installation in vehicle ................................................................. 74

Figure 43: Clutch release for human controlled launch .................................................... 75

Figure 44: Vehicle launch in simulation ........................................................................... 75

Figure 45: Simulation results for vehicle launch with stock Camaro gear ratios ............. 76

Figure 46: Simulation launch for stock ratios with aggressive controller ........................ 77

Figure 47: Simulated vehicle launch with EcoCAR gear ratios ....................................... 77

Figure 48: Clutch engagement during performance launch .............................................. 79

Figure 49: Clutch engagement logic in performance mode .............................................. 80

Figure 50: Breakdown of AMT control tasks ................................................................... 81

Figure 51: Torque pathways in series and parallel operation ........................................... 83

Figure 52: APP based clutch speed thresholds ................................................................. 85

Figure 53: BPP based clutch speed thresholds ................................................................. 87

Figure 54: Clutching logic and speed thresholds for engine only mode........................... 89

xiii
 Chapter 1: Introduction

1.1 Background

Increasingly strict government regulations pertaining to automotive fuel economy

and emissions, increasing awareness and willingness to curb pollution among consumers,

and the advancements in hybrid vehicle technology have all contributed to the increase in

the number of hybrid vehicles on the road [1]. While maintaining their efforts towards

increasing the efficiencies of conventional powertrains, most automotive manufacturers

are also pushing for increasing electrification of the automobile.

There are different types of hybrid vehicles, depending on their degree of

hybridization and powertrain architecture – ranging from mild hybrids (engine start-stop

and regenerative braking) to Extended Range Electric Vehicles (EREV). Hybrid vehicle

technology has penetrated the entire spectrum of automotive applications, from small cars

to construction equipment to Formula 1 cars.

To aid this effort and help develop the next generation of automotive engineers

through hands-on experience and exposure to hybrid vehicle technologies, the U.S.

Department of Energy (DOE) has sponsored Advanced Vehicle Technology Competitions

(AVTC) for almost 30 years starting from Methanol Marathon in 1988 to the ongoing

EcoCAR 3. AVTCs partner with automotive OEMs and are sponsored by automotive

1
suppliers, and thus provide opportunities for students by introducing them to advanced

propulsion and alternative fuel technologies.

1.2 EcoCAR 3

EcoCAR 3 is a 4- year competition sponsored by DOE and General Motors (GM),

and organized by Argonne National Laboratories (ANL), with 16 North American

universities transforming a stock conventional 2016 Camaro to a hybrid vehicle. The goal

is to re-engineer the Camaro to reduce its environmental impact while still retaining the

‘feel’ of a performance vehicle. The competition is structured to mimic an automotive

product cycle, with participants judged not only on the technical capabilities of the product,

but also their market research and consumer acceptability.

Figure 1: EcoCAR 3 vehicle development process

In Year 1, teams apply modeling tools to select a powertrain architecture that they

feel best supports their Vehicle Technical Specifications (VTS) and choose the powertrain

components from the choices available to them. Year 2 sees the maturation of their vehicle

2
simulation models with the chosen architecture, and the mechanical integration of the

vehicle, aiming at basic safe operation of the vehicle by the year end competition. Year 3

aims to have the vehicle operational in all modes, albeit not with the most sophisticated

feel, ready to complete all dynamic events at the year end competition, the most important

one being the Emissions and Energy Consumption (E&EC) event. Year 4, the final year of

the competition, aims at a marketable product, with the focus being on controls

optimization and consumer acceptability.

The Ohio State University (OSU) EcoCAR team’s vehicle has a Plug-In Hybrid

Electric Vehicle (PHEV) architecture, with the vehicle capable of operating in an Electric

Only (EV) mode, Engine Only mode, and series and parallel hybrid modes. It features an

economy mode tailored towards better energy consumption, and a performance mode,

making the vehicle more fun to drive. The vehicle architecture is shown in Figure 2, and

Table 1 lists the powertrain components and their specifications.

Figure 2: OSU EcoCAR 3 vehicle architecture

3
 Table 1: OSU EcoCAR 3 vehicle powertrain component specifications

Component Manufacturer / Model Specifications

BAS Motor Denso Integrated 32 kW, liquid cooled
Permanent Magnet Motor
BAS Inverter Rinehart PM100DX 100 kW, liquid cooled
Engine 2.0 L 4 Cyl DI E85 119 kW @ 6500 RPM
REM Motor Parker Hannifin IPM 112 kW, liquid cooled
REM Inverter Rinehart PM150DXR 150 kW, liquid cooled
Transmission TREMEC T5 5 - speed manual
transmission
Shifting Control Unit ASaP - Mastershift Electromechanical shifting
Clutch RAM Performance Single plate dry clutch
Clutch Control Unit FTE CP1 Electrohydraulic actuator
Energy Storage System A123 Systems 18.9 kWh, air cooled
HV Charger Brusa NLG 513 3.3 kW, air cooled
DC-DC Converter GM 2.2 kW, air cooled
Power Transfer Unit Team Developed 2.77 gear ratio
Rear Differential GM 2.77 gear ratio

The Internal Combustion Engine (ICE), and the Belted Alternator Starter (BAS)

constitute the front powertrain, and they are connected to the wheels through the clutch and

transmission, with the P3 Rear Electric Machine (REM) connected post transmission

through the Power Transfer Unit (PTU).

In Year 3 of the competition, the team developed the controls to make the vehicle

operate in all of its intended modes, and conducted validation of the controls software

4
through Model In the Loop (MIL), Hardware In the Loop (HIL) and Vehicle In the Loop

(VIL) testing. The controls focus in Year 3 was to have reliability over sophistication.

1.3 Thesis Objective

Ohio State’s EcoCAR vehicle has an Automated Manual Transmission (AMT) that

was developed by the team by automating a stock manual transmission. Year 3 involved

extensive work on the AMT system, from controls development to component testing and

in-vehicle validation. The thesis describes the modeling, testing, and controls development

of the AMT system during Year 3, and its operation during different hybrid vehicle

operation modes.

1.4 Thesis Organization

The subsequent portion of the thesis is organized as follows:

• Chapter 2 provides a brief overview on AMTs, their advantages in a hybrid

powertrain, and the research that has been done on AMT control.

• Chapter 3 provides an overview on the components of the AMT system in the

EcoCAR 3 vehicle and their operation.

• Chapter 4 describes the modeling approach for the individual subsystems.

• Chapter 5 details the component level testing done on the clutch and the

transmission subsystems to characterize their behaviour.

• Chapter 6 elaborates on the controls development process, deep dives into a few

key areas of the shifting logic, and shows in-vehicle results.

• Chapter 7 is the concluding portion highlighting the accomplishments and

challenges in key areas and the way forward for Year 4 of the competition.

5
 Chapter 2: Literature Review

2.1 Introduction

This chapter provides an overview of the background for the material presented in

this thesis. The different types of transmissions used in vehicles (conventional and hybrid),

and their advantages and disadvantages are discussed. Automated manual transmissions

are looked at in detail, focusing on their types, control approaches, drive quality perceived

by the consumer, and their applications in a hybrid vehicle architecture. The background

was important in the vehicle architecture selection as well as the controls development

process.

2.2 Transmissions Overview

A transmission forms an important power transfer point between the propulsion

systems and the wheels of an automobile. This enables the propulsive units to be paired

with the wheels through different gearing ratios based on the torque required, component

speed ranges, etc. In the case of an engine, it allows the engine to spin with the vehicle at

standstill to prevent stalling. Following are the prevalent transmission types-

2.2.1 Manual Transmission

Manual Transmissions (MT) have two parallel gear shafts – the main shaft and the

layshaft, connected with different gear ratios. The gears on the layshaft are rigidly attached

to it, while the ones on the mainshaft are ‘floating’. The gears on the mainshaft are coupled
6
to it using a shift fork, which is operated by the driver through the gear shifter. The

transmission is coupled to the engine through the clutch, which is operated by the driver

through a foot pedal. Most contemporary manual transmissions for road cars have 4 to 6

continuous synchromesh forward ratios and 1 reverse. Manual transmissions are relatively

simple, lighter, exclusively controlled by the driver, and more efficient than other types of

transmissions [2]. Table 2 shows some representative efficiencies.

Figure 3: Internal structure of a manual transmission [3]

Table 2: Typical gear efficiencies for a manual transmission [2]

Time in Gear Representative Current Production

Efficiency Efficiency Variation
Gear 1 8% 93.5% 92-96%
Gear 2 10% 92.0% 92-97%
Gear 3 21% 94.0% 93-97%
Gear 4 20% 97.4% 93-99%
Gear 5 41% 93.8% 92-97%

7
 The manual transmission has the highest efficiency of all the transmission types,

and the driver is always in control of the gearshifts. This makes the driver feel more

‘connected’ with the vehicle, but also relies on driver skill for smoothness during

gearshifting and vehicle launches.

2.2.2 Automatic Transmission

Automatic Transmissions (AT) consist of a pump, turbine, and a planetary gearset.

The driver does not have a separate clutch pedal and the driver cannot select a gear by

manipulating a stick. The vehicle’s Transmission Control Unit (TCU) selects a gear by

manipulating a system of clutches and brakes in the transmission to lock components of

the planetary gearset. There is no interruption of torque during the shifting process [4].

This type of transmission is very complex in structure which makes it heavy compared to

manual transmissions. Their efficiency is low compared to a manual transmission [2]. The

drive quality during shifting is better than that of a manual transmission and no driver effort

is required during launching and shifting.

Table 3: Typical gear efficiencies for an automatic transmission [2]

Time in Gear Representative Current Production

Efficiency Efficiency Variation
Gear 1 9% 70.3% 60-85%
Gear 2 4% 78.1% 60-90%
Gear 3 5% 86.0% 85-95%
Gear 4 32% 86.2% 85-95%
Gear 5 40% 88.7% 83-94%

New technologies like torque converter lock-up, increased gear ratios and optimization of

components have increased efficiencies of automatic transmissions [4].

8
2.2.3 Dual Clutch Transmission

The Dual Clutch Transmission (DCT) is structured like a manual transmission with

two layshafts, with the gears split between them. This transmission has two clutches, one

attached to each layshaft. Both layshafts are geared to the output shaft, and the layshaft

currently engaged to its clutch is the one transferring torque. The next gear ratio is already

selected on the other layshaft, and an upshift involves disengaging the first clutch and

engaging the second. The DCT control logic preselects the next gear anticipating an upshift

or a downshift, which makes sequential shifts very fast, but block shifts are comparatively

slower.

Figure 4: Schematic of DCT, ZF 6 speed AT [4]

2.2.4 Continuously Variable Transmission

There are two types of Continuously Variable Transmissions (CVT) – belt type and

toroidal type. Belt type CVTs have two discs connected by a belt, and a variator that

changes the distance between a pair of pulleys, changing the belt radius and thus offering

theoretically infinite transmission ratios. A toroidal CVT has two discs with inner surfaces

9
resembling a toroid, and a friction disc that transmits torque between the two discs.

Changing the angle of the disc causes it to contact different parts of the torus on each disc

leading to variable gear ratios. Belt type CVTs have overall efficiencies around 84.6% [2],

which are lower than an MT but allows the engine to be kept in its most efficient operating

zone. Toroidal CVTs have efficiencies around 91%. Belt driven CVTs have limited torque

transfer capability, and toroidal CVTs are heavy and require more manufacturing precision

than their AT counterparts [4].

2.2.5 Automated Manual Transmission

Automated manual transmissions are manual transmissions with the clutch

operation and gear shifting controlled by the TCU. The driver does not have a physical

clutch pedal and the gear shift lever is similar to an AT. Clutch and shift actuation can be

electromechanical or electrohydraulic, depending on the manufacturer/ supplier. The

transmission itself may be a modified manual transmission or it may use a stock unit. The

transmission efficiency is the same as a manual transmission, but the TCU selects the most

efficient gear ratio for an operating condition. Also, since vehicle launch and shifting are

electronically controlled, consistency is maintained and drive quality is not subject to driver

skill.

Compared to ATs, however, gear shifts see interruption of torque during shifting,

and vehicle launch quality depends on clutch control calibration since manual

transmissions do not have the ‘creep’ functionality inherent to ATs. Shift quality of AMTs

is usually rated poorer in comparison to ATs [5].

10
 AMT designs may be custom, where the internal structure of a manual transmission

is modified with automation as the goal. These modifications could include additional

clutches, engine brakes, synchronizer changes or torque assist motors [6]. These types of

transmissions are usually sold as complete units to OEMs. They usually represent the more

mature AMTs, but at the same time increase costs. The other design is ‘retrofit’

transmissions, where add-on units are added to existing manual transmissions, often

without modifications to the transmission internals. The shifting is slower compared to

custom AMTs, but they are more business-viable for small urban cars [6].

Figure 5: Classification of AMTs [6]

2.3 AMT Control

Taking into account the pros and cons of different transmission types, OSU

EcoCAR team decided to use an AMT in its vehicle architecture. The AMT was selected

for the following reasons:

i) To retain the high efficiency and low weight of an MT

11
 ii) To allow the supervisory controller to select the optimum gear depending on

operating conditions

iii) To leverage the hybrid architecture to improve drive quality compared to

conventional AMTs

During Year 1 of the competition, the team evaluated 4 different architectures for

performance and economy metrics [7] and selected an architecture with a 2.0 L engine and

a 5-speed manual transmission to be automated by the team. The team also leveraged its

previous experience in automating a manual transmission in EcoCAR 2. The important

aspects of a controller for the AMT are described below.

2.3.1 Vehicle Launch

In a conventional MT vehicle, the driver controls both the accelerator and clutch

pedal during a vehicle launch. In an AMT, control of the clutch is either electromechanical

or electrohydraulic. The aim of the clutch control is to achieve a clutch position in order to

follow a position trajectory commanded by the supervisory controller to achieve the vehicle

launch. The higher-level controller commands a suitable clutch engagement trajectory to

achieve a smooth launch, as is illustrated by an example shown in Figure 6. On receiving

a position command from the higher-level controller, the lower-level controller actuates

the motor to move the master piston to the desired position.

12
 Figure 6: Typical clutch engagement trajectory, actual launch at 30% APP [8]

This trajectory usually involves a quick release to the bite point, followed by a

slower release until the wheels get up to speed, and a quick release until full engagement

[8]. The clutch engagement trajectory also determines the gentleness of the engagement,

with a fast engagement needed for an aggressive launch. The TCU determines the

quickness of the clutch engagement based on the Accelerator Pedal Position (APP) [8].

Typical goals of vehicle launch are to minimize engagement time while providing a smooth

launch, avoiding engine stall and redlining during launch, and having minimum rollback

during a hill start [6].

Identification of the clutch bite point is an important step in the clutch engagement

trajectory. [9] describes a method for a clutch bite point estimation when pressure is the

control variable. [10] shows the nonlinear relation between the clutch diaphragm spring

displacement and stiffness, while [11] shows a similar relationship between throwout

bearing load and travel, as seen in Figure 7.

13
 Figure 7: Diaphragm spring force vs displacement [10], bearing load vs travel [11]

[12] illustrates how clutch torque changes with clutch wear, and [13] shows its dependency

on slip. These clutch characteristics were of great use when setting up the torque

transmissibility test rig.

2.3.2 Gearshift Sequence

The process of a gearshift (upshift or downshift) can be broadly sequenced as

releasing the throttle, disengaging the clutch, changing gears, engaging clutch, and

increasing throttle [14]. [12] further splits the process into the following phases:

i) Engaged – Clutch is completely engaged.

ii) Slipping opening – Clutch is slipping towards disengagement.

iii) Synchronization – Layshaft speed equals speed of next gear ratio.

iv) Go-to slipping – Engine speed starts to match layshaft speed.

v) Slipping closing – Clutch is slipping towards engagement.

Different controllers are used for each of these phases as they have different

objectives and variable states. The actual gearshifting phase, where the selector fork is

14
engaged with the desired gear, consists of getting a fork ‘out’ of gear, pushing the selector

fork until the synchronizer makes contact with the next gear, and completely engaging it

when speed match occurs. [6] makes use of position control until the synchronizer makes

contact, and switches over to force based control during speed match and dog engagement.

[15] makes use of sliding mode control and has the same phases for a gearshift. The control

relies on identification of the balk position along with the gear positions in the H shift

pattern, as shown in Figure 8.

Figure 8: Gearshift positions and control phases [15]

Position feedback is an important part of gearshift process. Controllers discussed

in [15] and [6] use analogue sensors to determine current position. Most approaches make

use of a selector position and shifter position to define the H pattern on an X-Y co-ordinate

system.

With increasing vehicle electrification, gearshift phases get modified to improve

drive quality, avoid shift shock and circumvent the torque interruption during shifting. In

15
general, hybrid vehicles with P1 motors can attain accurate speed matching on the engine

side since speed control can be better achieved with an electric motor than by manipulating

engine throttle. [16] describes a clutch-less shifting method for a hybrid vehicle

architecture with a P2 motor and a centrifugal clutch. The method performs a clutch-less

shift by shifting to neutral, uses the motor to match the engine speed to the expected

transmission input shaft speed of the next gear, and then shifts into the next gear.

To avoid complete torque interruption during shifting, [17] uses an actuated wet

clutch between the input shaft and layshaft as a torque assist mechanism. Hybrid vehicles

with a P3 motor can also use this motor as torque assist during gearshifts. The Zeroshift

AT has a modified shifter fork mechanism that eliminates torque interruption by shifting

in zero time [18].

2.3.3 Shift Scheduling and Drive Quality

Automating the manual transmission allows the TCU to be in control of selecting

the gear, and gear shift maps can be designed for economy and performance based on

vehicle speed and driver torque request, as is done for AT vehicles [5]. With hybrid

vehicles, the supervisory controller can request more torque from the motor and avoid a

downshift to keep the engine in its efficient zone [19].

Drive quality being an important part of consumer acceptability, smoothness of

clutch engagement and gearshifting should be paid attention to during controller

development and vehicle calibration. [5] and [20] performed a quantitative study of AMTs

by evaluating AMTs in different vehicle and comparing them with ATs. Different tests

such as creep, launch, hill starts and gearshifts were evaluated and on the whole, it was

16
found that consumers perceive AMTs to be worse than ATs in these aspects. Both papers

suggest that drivers tend to be more critical of AMTs than an MT due to the ‘lack of

involvement’ in the shifting process and vehicle jerks, which are perceived as normal when

driving an MT, are not tolerated when ‘someone else’ takes control. The main reasons are

the gearshifts taking the driver by surprise, the torque interruption during a gearshift and

longer gearshift time than an AT. Both studies found that in contrast with drivers, the

passengers viewed the AMT to be similar or better than the manual transmission.

Using an AMT in a hybrid vehicle architecture has the potential to overcome these

negative aspects by avoiding torque interruption during gear shifts and using speed

matching to minimize the shift shock.

17
 Chapter 3: Component Description

3.1 Introduction

This chapter describes the setup and operation of the components used in OSU

EcoCAR’s AMT system. The AMT system forms the junction between the front

powertrain and the rear powertrain, and can be used to couple or decouple them while

providing different torque transmission ratios. The AMT system can be split into the

clutching and shifting subsystems. Figure 9 shows the components, controllers, and

interfaces of the AMT system while Figure 10 shows their location in the vehicle.

Figure 9: Components in the AMT system

18
Clutch Clutch Transmission Shifting
Control Unit Control Unit

Figure 10: Location of AMT system components in OSU EcoCAR 3 vehicle

19
 The following are the two team-developed controllers that oversee the operation of

the vehicle:

i) Hybrid Supervisory Controller (HSC) – This is the vehicle supervisory

controller, and is referred to as ‘supervisory’ or ‘HSC’ throughout this thesis.

This controller is responsible for the entire vehicle operation, and it monitors

the operation of the individual component controllers. The HSC is responsible

for the energy management, vehicle mode operation, and high-level fault

detection and mitigation. The controller used is a dSPACE MicroAutoBox II

(MABx II).

ii) General Control Module (GCM) – This controller is primarily responsible for

the sensors and actuators on the vehicle since it has greater electrical ability and

can interface with switching relays and PWM signals as it has a higher current

capability than the HSC. The controller used is a Woodward SECM 112 control

module.

3.2 Clutching Subsystem

The clutching subsystem consists of the Clutch Control Unit (CCU), Concentric

Slave Cylinder (CSC), friction clutch, diaphragm springs and pressure plate. The CCU

consists of the master cylinder, piston, brushless motor, and the controller housed in a

single unit. The controller provides position control over the motor, thus controlling stroke

on the master cylinder side.

The CSC is connected to the CCU via hydraulic lines. It rests on the transmission

wall with the throwout bearing contacting diaphragm springs, and actuates the clutch just

20
like a standard hydraulic clutch, as seen in Figure 11. In the resting position, the motor

position is ‘0’, and the clutch is completely engaged. As an increasing motor position is

requested from the HSC, the CCU drives the motor to attain the position, pushing on the

piston, which in turn, pushes on the CSC to disengage the clutch.

Figure 11: Clutch actuation system

The supervisory level control for the clutch is based on the motor position and fluid

pressure reported by the CCU. The CCU is present on one of the vehicle’s 4 Controller

Area Networks (CAN) and sends out status messages and error messages, which the HSC

uses for its mitigation strategies.

3.3 Shifting Subsystem

The shifting subsystem consists of the Shifting Control Unit (SCU), shifting cables,

shifting adapter, and the 5-speed manual transmission. The SCU is an aftermarket unit for

manual transmissions, to replace the stock shifter with paddle shifters. As an aftermarket

unit, it is not CAN capable, and relies solely on electrical signals for operation.

The package includes a shifting adapter, which consists of a rod that replaces the

stock shifter. The adapter is controlled by two cables, one controlling the rotational motion
21
of the rod (the side-to-side movement of the stock shifter) and the other cable controls the

fore-aft motion of the rod (the front-back motion of the stock shifter). The cable controlling

the rotation is referred to as the selector cable, and the fore-aft cable is referred to as the

shifter cable throughout this thesis.

The cables are operated by two 12V motors controlled by the SCU. The motors are

individually attached to potentiometers via linkages. These potentiometers read the

position of the motors, and indirectly, the positon of the shifter rod. The position read by

the potentiometers is used by the SCU for feedback control when shifting, and is also used

by the GCM to determine the current gear. The lever arrangement is shown in Figure 12.

Figure 12: SCU actuator lever arrangement

The links of the lever arrangement shown in Figure 12 are:

A – Motor shaft to cable, shaft to potentiometer linkage

B – Motor link to potentiometer arm link

C – Potentiometer arm

D – Push pull cable to shifting adapter

22
 The same arrangement is in place for both the selector and shifter cables. The

difference lies between the motor forces. The selector motor experiences lesser force than

the shifter motor since the sideways motion (between 1-2, 3-4, etc) on a manual stick is

easier compared to the act of pushing it into gear.

As mentioned previously, the selector motor controls the rotational motion of the

shifting rod (sideways motion on a stick) while the shifter motor controls the fore-aft

motion of the rod (front-back motion on a stick). When the SCU is commanded to shift

from 2nd to 3rd, for example, the shifter motor first pulls the shifting rod from 2nd, the

selector motor then gets it from the 1-2 slot to 3-4 slot, and the shifter motor pushes it into

3rd . This process is illustrated in Figure 13, with the H shift pattern as reference.

Figure 13: Actuation sequence for 2-3 upshift

During initial programming of the SCU, the cables were first disconnected from the

motor arms, the transmission was manually moved into gear, and the motors were
23
individually commanded into position until the cable disconnected could fit snugly in

place. These positions were calibrated as row-column positions into the SCU through

Electrically Erasable Programmable Read Only Memory (EEPROM), and the

potentiometer readings were considered while setting the boundaries on the GCM to

determine motor positions.

The EEPROM interface allowed the shift positions to be programmed, and shifts

could be commanded individually through EEPROM. This was done for the initial

sequential testing, to tweak the position values a bit to allow for the variance of the SCU

control. The shifting package, being designed for aftermarket use, has optional protection

features in it, that can be set through EEPROM. These are downshift protection (does not

allow downshifts at higher engine speeds), reverse protection (does not shift into reverse

unless speed is low), clutch safety (does not shift unless clutch pedal is pressed).

The gearshift commands are supposed to be received by the SCU through two

steering column mounted shift paddles – one for upshifts and one for downshifts. In the

EcoCAR vehicle, these two wires are switched by the GCM, emulating the paddle shift

functionality. GCM Low Side Output (LSO) drivers are used that ground these wires,

complete the circuit in the SCU, and request an upshift/ downshift.

The ‘Clutch Safety’ feature of the SCU is enabled in the vehicle. This feature causes

the SCU to wait for a signal indicating that the clutch pedal is depressed. This signal is sent

out by the GCM, after looking at the clutch status sent out by the HSC over CAN. While

this feature in itself is not necessary, since the HSC and GCM both look at clutch status

before requesting a shift, it allows block shifts to be performed. This is accomplished by

24
requesting multiple upshift/ downshift blips, depending on the increments needed, and then

sending the clutch signal. The SCU counts the number of shifts before the clutch switch

signal, and shifts accordingly.

As mentioned before, the SCU uses the potentiometers to determine the position of

the selector and shifter arms, and detects the gear based on these positions. The SCU sends

out the indicated gear over a one-wire network, meant for a digital gear indicator in the

cockpit for aftermarket applications. This protocol, however, cannot be utilized by any

team developed controller, and as a result, probes had to be inserted into the rotary

potentiometers and sent as analogue inputs to the GCM. The GCM thus determines the

gear independent of the SCU.

The SCU has another parameter, ‘stall timeout’. This parameter is the time the SCU

keeps trying to power the motor for if the selector/ shifter does not reach the desired

position, before retreating. This parameter is meant to stop trying to shift in case of a

blocked shift. The parameter had to be updated since it was found during testing that the

default timeout was less than the time it took for the synchronizers in the transmission to

spool up to make the shift possible. Using the default timeout worked in static testing, but

would ‘time out’ when shifting at speed. The stall timeout threshold was increased to allow

for synchronization time.

25
 Chapter 4: Modeling

4.1 Introduction

This chapter introduces the full-vehicle model, the clutch and transmission

subsystems modeled for full-vehicle and standalone simulations, and the softECU – plant

structure followed by the various component models.

The team’s full-vehicle model is called EcoSIM 3, and is developed in the

MATLAB/ Simulink environment. Its function is to model the different vehicle

components and their controllers to provide a platform to test the vehicle supervisory

controller logic. This supports the team’s core philosophy of Model Based Design (MBD),

allowing supervisory logic testing before availability of the vehicle prototype. EcoSIM 3

is an energy based model that has a PI controller to act as the ‘driver’, generating

accelerator pedal and brake pedal signals to provide torque request to the vehicle

supervisory controller. The full-vehicle model helps in estimation of energy consumption

of the vehicle and testing out different supervisory energy management strategies.

EcoSIM 3 is structured in a modular fashion, with the full-vehicle model being split

up into three sections: Supervisory controller, GCM, and powertrain model, as shown in

Figure 14. This structure allows for easy splitting when validating the controller through

different in the loop (xIL) testing phases.

26
 Figure 14: Modular structure of EcoSIM 3

The powertrain section of EcoSIM 3 consists of the energy storage components,

propulsive components and the vehicle road load model. The structure of the model is in

sync with the actual torque flow, allowing for a more intuitive understanding of the model

structure. Each component is further split into a softECU, which models the component

controller, and the plant, which models the physical side. The softECU subsystem

communicates with the supervisory controller and the plant sensors. This makes it possible

to emulate the communication as it happens on vehicle. The component level structure is

shown in Figure 15.

27
 Inputs from Outputs to
other SoftECU other
controllers controllers
Input Algorithm Output
Plant Model To Plants
From Plants Actuator Component Sensor
Models Models Models

Sensor Signals

Figure 15: Component level structure of EcoSIM 3

4.2 Modeling Approach

The OSU EcoCAR team has adopted a Model Based Design approach based on the

Systems Engineering V- diagram, shown in Figure 16. This approach focuses on

development of a model that allows controller testing to start much before a prototype is

available.

Define Complete
Requirements Integration & Test
Continuous
Engineering
System-Level System-Level
Specifications Integration & Test

Subsystem Subsystem
Design Integration & Test

Subsystem
Implementation

Time

Figure 16: V diagram for MBD process

28
 The process begins with defining vehicle requirements. Requirements are defined

based on component documentation, simulations, past experience, and are updated as new

information is available. The vehicle model and the controller logic both go through the

development process in parallel. Preliminary models are developed based on requirements

and component documentation. Preliminary testing and controls development can be

started at this stage since the model developed from component documentation is sufficient

to give an estimate of the energy consumption and general vehicle behaviour. The controls

logic then goes through the different ‘In-the-loop’ xIL phases, undergoing refinement in

each stage.

Component level testing is one of the most beneficial test phases, since it helps

realize the interface between the supervisory controller and the component controller.

Component handshakes, signal behaviour, component behaviour can be observed and

implemented into the softECU and plant models. Actual component behaviour and

response times also help to update the requirements to make them more realistic.

The controller and the model jump back and forth between the two arms of the V

process, and this results in continuous model updates. The modular structure of EcoSIM

goes a long way in easing the xIL transition and thus streamlines the process of regression

testing.

The models have the following goals to support controls testing:

i) Suitable I/O fidelity to support requirements testing

ii) Suitable I/O nature to match test environment

29
 iii) Implementation of softECU features and signals needed to test fault detection

and mitigation strategies

The component model development process is summarized in Figure 17.

Figure 17: Component model development summary

Selection of the softECU I/Os to be modeled is an important part of the softECU

development process. A component controller transmits a variety of signals on the CAN

bus, from heartbeats and physical values to Diagnostic Trouble Codes (DTC) and error

signals. It is important to identify the signals that will be used by the supervisory controller

and correctly model them to mimic the real process. While all the signals can be modeled,

it is not necessary and only adds to development time without helping in controls

development. Table 4 shows an example of softECU I/O selection for the CCU softECU

development.

30
 The Clutch Control Unit sends out a wide range of signals over CAN, ranging from

physical parameters like piston position, line pressure, motor torque, current, etc. to

diagnostic signals. The supervisory controller makes use of only a part of these signals for

component operation and fault detection. It is enough to model just these signals and model

them accurately.

Table 4: CCU softECU I/O selection

Total I/Os Required In Vehicle Modeled

Power 2 2 2 2
Commands 11 3 3 3
Heartbeat 2 1 1 1
Physical Measurements 22 2 2 2
Component Status/ Errors 11 8 8 8
Component version 5 0 0 0
Reserved 9 0 0 0

4.3 Clutching Subsystem

The clutching subsystem used in the vehicle consists of the Clutch Control Unit,

the Electrohydraulic Actuator, the hydraulic lines, the Concentric Slave Cylinder, and the

disc friction clutch.

The CCU is CAN controlled and is present on the Front Powertrain CAN

(FPTCAN), which is the CAN network consisting of the Front Powertrain component

controllers, the GCM, and the supervisory controller. The CCU controls the

electrohydraulic actuator, which is a brushless motor that drives a recirculating ball spindle

that actuates a hydraulic piston to create pressure and displace fluid.

31
 The CSC is used to apply pressure on the fingers to disengage the clutch, and is

from a stock automotive application. The components used in the clutching subsystem are

listed in Table 5.

Table 5: Clutching subsystem components

Component Interface
Clutch Control Unit CAN , 500 kbaud
Electrohydraulic
Brushless motor
Actuator
Slave Cylinder Hydraulic
Friction Clutch Mechanical

The softECU for the CCU is modeled to emulate the startup and shutdown

procedure, along with normal component operation and component level fault detection.

A well-modeled softECU has enough I/Os and the appropriate level of fidelity so that the

supervisory controller interfaces with the CCU as it would on the vehicle, and the CCU

interfaces with the plant sensors and actuators.

The softECU is modeled to represent CCU operation to achieve the desired

functionality and ensure the appropriate level of component- level fault handling. The

various states have been summarized in Figure 18, while the CCU functions during those

states are detailed in Table 6. All the modes and their associated functionalities have been

modeled in the softECU, with the exception of the ‘Remove Air Mode’. The team does not

use the air removal mode provided by the CCU. Table 7 details the signals modeled in the

softECU.

32
 Figure 18: CCU operation states

Table 6: CCU operation states

CCU State Transition Action

Secure Mode On powerup Clutch engaged
Operation Mode Mode request Obey position request
Clutch engaged, error
Fault Component fault
message sent on CAN
Shutdown Sleep request Proper shutdown
Remove Air Mode Mode request Pump piston to remove air

Table 7: CCU I/O signals

Signal Nature Function

Powerup Electrical Input Component power
Wake Electrical Input Controller wake
MU02_st_ModeReq CAN Input Mode request
MU02_pos_Des CAN Input Position request
MU01_SleepAckn CAN Input Shutdown request
CP01_Counter CAN Output Heartbeat
CP01_stMode CAN Output Current mode
CP01_pos_Curr CAN Output Current position
CP01_p_Curr CAN Output Current pressure
CP01_errX CAN Output Error
33
 The signals are modeled as per their nature on the HIL simulator e.g. powerup and

wake signals are physical electrical signals, while the CAN signals are sent over HIL CAN

using component DataBase Container (DBC) files.

Emulating mode and fault functionality has the advantage of being able to simulate

component level faults during MIL and HIL testing, allowing the supervisory controller to

receive the signals it would receive on the vehicle, and thus perceive a fault in a realistic

manner.

The clutch disc is modeled using SimDriveline Disk Friction Clutch block available

in Simscape. This block has two inputs – one for the input torque and one for the normal

pressure. The actual line pressure is modeled using a lookup table that outputs the line

pressure based on the piston position. A standalone Simulink model was used to model the

piston position – line pressure relationship. This model was made using SimHydraulics

blocks in order to better capture fluid material properties. The standalone model is also

capable of modeling air in the clutch lines.

Imperfect bleeding of the clutch lines during installation leads to trapped air in the

lines. This leads to more piston stroke being required to generate the pressure to be able to

push the pressure plates and disengage the clutch. More air in the system leads to the

possibility of not being able to achieve clutch disengagement even with complete piston

movement.

The model was used to generate position – pressure curves for varying amounts of

air in the lines, and was compared to the curves obtained during bench testing with different

bleeding procedures.
34
 Figure 19: Motor position – line pressure behaviour with varying air amounts

The data trend from the model is the same as the trend observed during testing, with

the line pressure rising slowly at first as air gets compressed. The standalone model was

useful to design the ‘Clutch Initialization Procedure’, which will be dealt with in the later

sections.

The standalone model was only used to model air in the system. It was not included

in EcoSIM 3 since the fast pressure dynamics would necessitate a stiff solver with variable

step time, greatly increasing simulation time, and making it unsuitable for HIL simulations.

In EcoSIM 3, electrohydraulic actuator position – line pressure – transmissible torque

behaviour observed during bench testing was implemented using lookup tables. The black

box approach helped reduce simulation time while still maintaining the motor position –

35
clutch torque relationship. This approach does not, in any way, compromise the fault

insertion capability of the model. The model structure is summarized in Figure 20.

Figure 20: Clutch subsystem model summary

4.4 Shifting Subsystem

The shifting subsystem consists of the Shifting Control Unit, actuator motors,

shifting cables, and the 5- speed manual transmission. The SCU accepts shift requests from

the GCM, and controls the shifter and selector motors in order to move the shift forks and

select the desired gear.

The SCU does not have CAN capability, and functions completely on digital inputs,

with upshift and downshift ‘blips’ being used to request the desired gear. Since it was meant

36
to be an aftermarket product to add ‘paddle shift’ functionality to a stock manual

transmission, it has limited diagnostic capabilities. The SCU uses potentiometers attached

to the motor arm to determine the shifter / selector position. The GCM is probed into these

potentiometers and is thus able to detect shifter / selector arm positions. The potentiometer

signals are the only feedback signals available from the SCU. The SCU I/O capabilities are

listed in Table 8.

Table 8: SCU I/O signals

Signal Nature Function

Powerup Electrical Input Component power
Electrical blips to indicate gear
Upshift_Sw Electrical Input
increments
Electrical blips to indicate gear
Downshift_Sw Electrical Input
decrements
Electrical blip to indicate clutch
Clutch_Sw Electrical Input
status
Analogue signal indicating
ShifterPos Electrical Output
shifter pot position
Analogue signal indicating
SelectorPos Electrical Output
selector pot position

These signals are implemented as electrical signals on the HIL. Certain adaptations

were made in the HIL wiring harness to implement the Switch signals. The GCM operates

the switch signals through its LSO drivers. An LSO needs to have a pull-up resistor

somewhere in the circuit. The SCU has an internal pull-up that can be selected through

EEPROM. The schematic in Figure 21 shows the external pull-up resistors that had to be

added to the HIL simulator wiring harness to have proper inputs going in to the SCU

softECU.
37
 Figure 21: Addition of pull-up resistors for HIL simulator wiring harness

The internal operating states of the SCU are unknown since the component

documentation does not mention operating states and no such diagnostic information is

available through serial interface. This resulted in the softECU being modeled with just

two basic operating states – ‘Off’ and ‘Normal Operation’, depending on the state of the

SCU powerup. The states are summarized in Table 9.

Table 9: SCU operating states

SCU State Transition Action

Shifting not available,
Off Initial / power off
sensors powered off
Obeys shift requests,
Operation Mode Power on
sensors powered on

The softECU counts the gear increments / decrements requested using Simulink

Counter block. The shift counting operation was implemented on the softECU to mimic

the behaviour of the SCU. For example, if the current gear is 2nd and 5 upshift blips are

38
sent, the SCU shifts into 5th gear. Similarly, if 4 downshift blips are requested, the SCU

shifts into neutral. The SCU had its EEPROM parameters set so that it would include

neutral in the counting sequence, and reverse gear functionality was disabled.

The clutch switch functionality was also implemented in the softECU to allow

block shifts to happen. This is done by counting the upshift/ downshift blips until the

Clutch_Sw signal is true, then shifting to the appropriate gear.

The SCU affords many protective functionalities, such as overspeed protection,

where it does not allow a downshift when there is a risk of overspeeding the engine. These

functionalities were disabled through EEPROM, since they are inherently taken care of by

the supervisory controller. Consequently, they were not included in the softECU logic.

The transmission is modeled using ‘Double- Sided Synchronizer’ blocks available

in SimDriveline. The block represents a double-sided synchronizer with two back-to-back

dog clutches, two back-to-back cone clutches, and one translational detent. Translating the

shift linkage in either direction causes it to engage with the cone clutch- dog clutch on that

side, causing it to transmit torque input through the clutch pair. This mechanism simulates

the behaviour of the actual synchromesh gearbox, with the shifting linkage translation

being similar to moving a shift fork in the transmission. The translational detent represents

the force required to overcome the speed difference of the cone clutches and engage the

dog clutch. This also serves to model the physical force required to move the shift fork and

‘select’ a gear.

The double-sided synchronizer block can model one pair of gears on the same shift

fork, and hence three double-sided synchronizers were used to model gear pairs 1-2, 3-4,

39
and 5-R, representing the arrangement on H pattern shifts. It must be mentioned that in an

actual manual transmission, reverse gear does not have synchromesh, and is often selected

by sliding the idler gear into place. Often, 5th and R are not present on the same shift fork.

For the sake of simplicity, the model has 5 and R on the same double-sided synchronizers,

and no idler gear is modeled. The arrangement is shown in Figure 22.

Figure 22: Transmission model with double-sided synchronizers

The gear ratios were set using Simscape ‘Simple Gear’ blocks added on the input

to each double-sided synchronizer block. The constant efficiency setting was used and an

efficiency of 90% was used for all gears.

The double-sided synchronizers do not model inertia. A lumped layshaft inertia was

modeled on the transmission input. The lumped inertia assumption is valid since all gears

40
move along with the transmission shaft and countershaft, and the inertia remains constant.

Thermal effects were not modeled for the transmission.

The shifting mechanism was modeled using a PI controlled actuator to apply force

to cause movement on the shift linkage, as shown in Figure 23. The PI can have three target

positions for each double-sided synchronizer as shown in Table 10.

Table 10: Shift linkage positions

Shift Linkage Position Torque connection

Center No torque transfer
Torque transfer from left conical clutch i.e.
Left
gear ratio corresponding to left input HA
Torque transfer from right conical clutch i.e.
Right
gear ratio corresponding to left input HB

Figure 23: Gear shift actuating mechanism

41
 The double-sided synchronizer block outputs the translational position of the shift

linkage as well, making feedback control possible, and allowing potentiometer sensor

models to be based off these positions. The softECU actuates the proper shift linkages

based on the gear shift desired through the PI controller. An upshift from 2nd to 3rd gear,

for example, would require pushing the linkage on 1-2 synchro from ‘right’ to ‘center’ (out

of 2nd into neutral) and then pushing the 3-4 synchro shift linkage from ‘center’ to ‘left’

(from neutral to 3rd). Simscape linkage positions are shown in Figure 24.

Figure 24: Shift linkage positions for 2-3 upshift

42
 The PI parameters were tuned to obtain the same shift times as those taken by the

shifting unit-transmission combination during bench testing. A comparison is shown for

the 3rd to 4th upshift in Figure 25.

Figure 25: Shift time comparison for 3-4 upshift

The sensor models convert the shift linkage positions reported by the Simscape

blocks into a selector potentiometer voltage and a shifter potentiometer voltage. The

voltage values are selected based on the values observed from component testing. As in

the actual case, the selector voltage is based on which double sided synchronizer is

selected, and the shifter voltage is based on the shift linkage position. Figure 25 shows

these voltages for the 3-4 upshift.

The sensor voltages are used by the GCM to determine the selector and shifter arm

positions and thus recognize the current gear. These signals are implemented as analogue

43
electrical signals on the HIL simulator. A summary of the entire shifting process is shown

in Figure 26.

Figure 26: Shifting subsystem model summary

44
 Chapter 5: Component Level Testing

5.1 Introduction

This chapter describes the testing done at a component level for both the clutching

and the shifting subsystems. The team has performed component level testing on the major

powertrain components, to validate their operation and controls before integrating them on

the vehicle. The purpose of the component level testing in the case of the clutch and the

transmission was to get the controller handshakes down correctly and characterize

component behaviour to increase the fidelity of the simulation models.

5.2 Controller I/O Testing

Component documentation usually describes the component controller operation in

detail and is enough to give a basic understanding of the controller startup procedures and

fault handling capabilities. Testing the controllers is essential to validate the controller

handshakes and to make sure the signals being sent out by the team-developed controllers

are in the manner the component controller expects to receive them. This process usually

involves providing 12V power supply and wake signals to the ECUs and trying to

communicate with them over a suitable communication channel (CAN), then sending out

the handshake signals manually one at a time until the controller reaches its operational

state. Testing out ECU I/Os before bench testing ensures the controller is ready for the

bench testing.
45
 The CCU was tested by connecting it to CANoe, powering it up, and recognizing

the version of the CAN DBC file flashed on it, since the documentation provided by the

supplier had three different dbc files. The CCU was then connected to the HIL simulator,

and operation in different modes was verified. Piston movement at this stage was executed

slowly since the CCU was standalone without any load on it.

The SCU was tested for I/O compatibility in a similar fashion, connecting the

Upshift, Downshift, ClutchSwitch, and two potentiometer sensor wires to the GCM, and

commanding shifts from the GCM. Sequential shifting and block shifting (without shifter

cables attached) was tested to verify proper signal generation. The smallest time for which

a blip could be sent out so that it could be detected by the SCU was also determined. It was

necessary that the blip time be small enough since that added to the required shifting time.

For example, to downshift from 5th to 2nd requires 3 downshift blips. Setting the blip ‘ON’

time to be 0.5 sec requires a blip generation time of 3 sec, adding a lot of time to the shifting

process. Making the blip time too low, on the other hand, would confuse the SCU, making

it wrongly count the blips and shift into the wrong gear. The smallest time for shift

detection was found to be 0.07 sec, and it was set to be 0.1 sec in the logic for robustness.

After the SCU and CCU were individually tested and the I/O interface was

established, both were connected to their respective controllers, and the entire shifting

logic, elaborated in the next chapter, was executed to check their behaviour in real time,

set the appropriate time thresholds to account for signal noise, delays, etc. This testing,

while not representative of in-vehicle behaviour, involved actual controllers and resulted

46
in a first-cut validation of the handshakes and operation. Figure 27 shows the setup used in

this I/O testing.

Figure 27: SCU I/O testing setup

5.3 Shifting Testing

The transmission was tested on a test rig consisting of the 5-speed transmission and

an electric motor attached to the input shaft. The purpose of the motor was to provide

motion to the input shaft to check shifting while providing rotation to the components to

smoothen the shifts. The transmission was connected via the shifting cables to the SCU,

which was, in turn, connected to the GCM, the controller responsible for the shifting

operation. The purpose of this test rig was to conduct an accelerated shift testing to identify

problematic shifts patterns, if any, and set the correct position and temporal limits for

potentiometer-position based gear detection by the GCM.

47
 The gear position co-ordinates were first mapped manually by commanding the

SCU into different positions through the serial interface. These positions were found to be

slightly different than the mapping done without the GCM in the loop, since the GCM

connected to the potentiometers introduced external resistance in the circuit. Calibrating

the gear position co-ordinates requires some iterations to get them ‘just right’ (make sure

the variance on the motor position would still get the shifting rod into the correct slot for a

successful shift).

Accelerated shift testing was done by feeding automated gear requests to the GCM

and having it generate the corresponding blips as it would on the vehicle. The shifts were

sent out both in sequential (N-1-2-3-4-5) and block (2-4, 5-2, etc.) patterns. The following

issues were encountered and resolved during testing:

i) Motor Issues – During initial stages of testing, the shifter arm on the motor kept

shearing off, resulting in a slotted arm being put in. After about 1500 shifts, the

shifter motor itself broke, with the nylon teeth inside the motor being stripped off.

This resulted in new motors with metal teeth and higher torque rating being used

on the shifting unit.

ii) Slack in joints – The stock shifting package had the cables connect to the studs on

the shifting unit and shifting adapters by quick disconnect versions of rod ends.

This arrangement inherently has play at the ends. The joint play, coupled with the

play in the shifting adapter, was enough to cause the unit to miss shifts occasionally.

This slack was confirmed to be the cause since when the positions for the selector

motor were noted after a failed shift, and they were the same positions programmed

48
 into the SCU. This led to modifications being made in the linkages. New rod ends

were installed and the play in the system was reduced. Along with the mechanical

modifications, the ‘slack compensation’ feature of the SCU was used. This feature

is used to adjust for cable slack and makes the motors overshoot their target position

by a programmed amount. These two modifications made the shifting more robust

and reduced the variance caused by slack.

iii) Potentiometer failures – It was noticed a few times that during component powerup,

the selector potentiometer would read invalid values, both on the serial interface

and the actual potentiometer voltage as measured by a multimeter. The failure

manifested as a ‘pull-up to supply’, with the potentiometer reading its extreme

position and the voltage reading the supply voltage. This voltage would remain

pulled up even if the potentiometer arm was moved. The SCU does not look for

sensor validity, and read the potentiometer as being at the extreme end of its travel.

In such a condition, if the SCU is commanded to shift, it tries to actuate the motor

to move the arm to the other end, but has no position feedback since the

potentiometer reads a constant voltage. This results in the motor arm moving until

it is physically blocked either by the transmission or the linkage constraints. This

resulted in implementation of potentiometer validity detection on the GCM and

HSC, with shifts not requested in case of invalid readings from the potentiometers.

This issue was taken care of by re-soldering the potentiometer connections. This

issue has not been observed on the vehicle at all.

49
 The shift testing had around 4000 shifts being put on the test rig, validating the

controller handshakes and multiple shifts through all gear combinations except reverse.

While reverse gear was initially programmed through EEPROM, and tested on the rig, it

was disabled on the SCU since vehicle travel in reverse was decided to be done through

the REM alone. A sample plot of the selector-shifter potentiometer read values is shown in

Figure 28. This plot was useful to observe the variance of the positions, both due to actual

position variance, as well as the electrical noise.

Figure 28: Potentiometer reported values on shifting test rig

This position variance helped to set the limit threshold on gear position recognition

to mark a successful shift by the SCU. The thresholds thus set were neither too restrictive

nor too lenient, and avoided overlap between adjacent positions.

50
5.4 Clutch Testing

After testing the shifting subsystem, the transmission was used on a different test

rig, this time with the clutch and Dual Mass Flywheel (DMF) attached to it. The aim of this

testing was to have the clutching subsystem installed as it would be on vehicle, validate

clutch operation, and characterize the torque transmissibility of the clutch at different

positions from complete engagement to complete disengagement. This behaviour was then

used to fine tune the clutching strategies and increase the fidelity of the plant models.

The transmission was supported by the test rig, and the dual mass flywheel was

bolted to the test rig frame, as shown in Figure 29. This had the effect of locking the

flywheel in place to serve as reference to observe clutch slip. The transmission was set in

4th gear, since it had a 1:1 gear ratio and input torque was thus same as the output torque.

The output shaft of the transmission was free to rotate. This was attached to a yoke with a

hexagonal bolt (the torque application point), and a torque wrench was used to apply torque

at this point to record the torque at which the clutch began to slip. On the vehicle, the engine

output shaft spins clockwise (seen from the front), and so does the transmission output

shaft. The engine thus sees the road load torque acting in the opposite direction (clockwise

seen from the back). The torque wrench was thus used to apply torque in the clockwise

direction from the output shaft. The 1:1 direct ratio meant the torque recorded by the

wrench was the slipping torque seen by the clutch.

51
 Figure 29: Torque transmissibility test rig

The test rig was first used to observe the CCU piston position – line pressure

behaviour, to set the overpressure protection limits and design the pressure based control

that was employed for the Clutch Initialization Procedure (CIP). The CIP is dealt with in

detail in the next chapter. The data obtained from these tests was also used to develop the

models mentioned in the previous chapter. The pressure trends were obtained each time

the installation and bleeding was carried out. At each motor position, the line pressure and

displacement on the CSC were recorded. The pressure trends are shown in Figure 30.

52
 Figure 30: Motor position – line pressure plots on different setups

The procedure followed for the torque transmissibility test was to command the

motor to various positions in steps from completely disengaged (the output shaft could be

spun freely by hand) to completely engaged (zero position on motor). Torque was applied

on the wrench until it began to slip – this was recorded to be the peak torque, and further

torque was applied to keep the clutch slipping steadily- this was recorded as the slipping

torque. An average of 3 to 4 readings was taken to get both the peak torque and the slipping

torque corresponding to the clutch position. The position- torque plot for the test rig is

shown in Figure 31.

53
 Figure 31: Position – torque and pressure – torque characteristics

In Figure 31 it is seen from the position – torque plot that as the clutch is released

(going right to left on the x axis) there is no torque transmitted during the initial travel, then

it increases almost linearly and plateaus to the maximum transmissible torque around 250

Nm. This behaviour is similar to what is experienced with a conventional clutch pedal. The

bite point is usually around 60% to 80% disengagement. If the transmissible torque is

plotted against percent engagement in Figure 32, a more intuitive idea is obtained.

The vehicle launch bite point shown in Figure 32 is the clutch position required for

initial vehicle movement during an engine only launch. To move the EcoCAR 3 vehicle

from standstill, a wheel torque of about 180 Nm is required. This translates to about 20 Nm

of torque on the engine side (and roughly the same at the clutch side), which is attained at

around 30% engagement (which corresponds to 70% disengagement).

54
Figure 32: Clutch transmissible torque plotted against percent engagement

55
 Chapter 6: Control

6.1 Introduction

This chapter explains the control logic and strategies used to handle the gear

shifting process. Owing to the different vehicle operating modes available due to the hybrid

nature of the vehicle, strategies differ according to the operation mode of the vehicle.

Depending on the operation mode and component speeds, different clutch engagement

strategies are employed to suit differing scenarios from vehicle launch to a series-parallel

mode transition.

6.2 Shifting Logic Overview

The overall shifting process is controlled by the supervisory controller, and the

logic is situated in the ‘Propulsion Actuators’ section of the supervisory code structure. It

is placed after the ‘Mode Operation’ logic and before the ‘Torque Security’ logic. This

placement allows the shifting logic to access component status through CAN and accept

gearshift commands and clutch engagement mode requests from the Mode operation

section.

The ‘AMT Actuators’ section is further broken down in to three main sections:

i) Overall Shifting Logic

ii) Clutch Initialization

iii) Clutch Engagement Control

56
 The Shifting Logic section deals with the complete gear shifting procedure, from

receiving a gear change request to handling component torques, performing clutch

disengagements and requesting gear change from the GCM. The process is summarized in

Figure 33.

Figure 33: Overall shifting logic in HSC

The important steps involved are:

1) Engine torque cutoff – Accelerator pedal signal to the Engine Control Module

(ECM) is set to zero, cutting off torque coming into the clutch.

57
2) Clutch disengagement – After engine torque is cutoff, a disengagement request is

sent to the later part of the logic, where it gets converted into a position request for

the CCU.

3) Gear change – The HSC receives the clutch piston position over CAN. Once the

clutch is confirmed to be in the disengaged state, the HSC sends out the gear request

as well as the clutch status to the GCM over CAN. The GCM performs the gear

shift, as explained in the later sections, and sends out the current gear and

‘Shift_Complete’ signal over CAN.

4) Speed match (optional) – Once the transmission is in the desired gear, the HSC

estimates the transmission input shaft speed for the new gear ratio based on the

REM speed. This speed is set as the target speed for the PI controller that requests

torque from the BAS to change the engine speed so it matches the transmission

input speed. The speed matching is done if the BAS is in the proper operational

state and the engine speed is not high enough. This part of the logic has not been

validated on the vehicle. The speed match is said to be complete when the engine

speed and transmission input speed difference has been less than ~100 RPM for

more than a certain time.

5) Re-engage clutch – In this phase, a clutch engagement request is sent to the ‘Clutch

Engagement Control’ section of the algorithm. The section engages the clutch using

an engagement strategy suitable for the vehicle speed and the speed difference

between the engine and the transmission.

58
 6) Allow engine torque – After clutch engagement is complete, driver torque request

is again sent out to the engine, after being rate limited depending on the vehicle

mode in operation.

The speed matching process using the BAS is currently not implemented on the vehicle,

and the different phases of shifting process are shown for 3-4 upshift in Figure 34.

Figure 34: Shifting process events for a 3-4 upshift in engine only mode

6.3 GCM Shifting Logic

The GCM controls the actual shifting process through the SCU. It is connected

electrically to the SCU and sends out the Upshift/ Downshift blips to the SCU through its

LSO drivers. The HSC requests a gear shift to the GCM through CAN and the GCM sends

the shifting status and SCU status to the HSC. Figure 35 shows the shifting logic.

59
 Figure 35: Shifting logic in GCM

Following are the important steps in the execution of a gearshift by the GCM:

1) Blip generation – The GCM compares the new gear request with the current gear,

determines the incremental difference, and generates the corresponding number of

blips. For example, a shift from 2nd to 4th will generate 2 upshift blips while a shift

from 2nd to 1st will generate 1 downshift blip. The shortest blip time offering reliable

shifting as observed during component testing was 0.07 sec. To account for larger

transport delays and in-vehicle noise, this blip time was set to 0.1 sec. After gear

shift blips are generated, the ‘Clutch_Sw’ blip is generated. This is an indication to

the SCU that the clutch is disengaged and it can start actuating the shift cables. An

example of blip generation for an upshift and a downshift is shown in Figure 36.

60
 Figure 36: Blip generation by GCM

2) Shift reporting – The GCM monitors the selector and shifter potentiometer voltages

to determine the current gear. If the actual gear does not match the requested gear

within a time threshold, the GCM tries to resolve the failed shift. A failed shift

could either be a shift into an incorrect gear, or a mechanical blockage causing the

shifter/ selector to get stuck at an odd location.

3) Failed Shift Handling (FSH) – In case of a failed shift, the GCM asks for a shift to

neutral, and then tries to shift into desired gear again. After three failed attempts, it

flags a fault and declares a failed shift. Temporary shift blocks are resolved by this

retry method, while something more serious warrants further investigation and the

number of tries are limited to three. Another method of handling failed shifts was

to shift into neutral, shift into 3rd or 4th to realign the dog clutch teeth and then try

61
 to shift again. Most of the failed shifts were found going into 1st gear, as is often

the case with manual transmissions.

6.4 Clutch Initialization Procedure

The CCU initialization handshake, position determination, and fault handling is

handled in this section of the logic. CCU initialization handshake occurs during vehicle

start up, and involves powering up the CCU, receiving status messages on CAN and

requesting a transition to ‘Operation Mode’. In case of a fault reported by the CCU, the

mode requested is ‘Secure Mode’, and CCU failure section of the supervisory failure

detection and mitigation strategies is activated. During vehicle shutdown, the clutch is first

engaged and sleep signal is sent to get the CCU into shutdown mode.

Figure 37: CCU startup and shutdown handshakes

The entire clutch engagement-disengagement strategy is developed in terms of

normalized clutch position, where 0 represents fully engaged position and 100 represents

62
fully disengaged. Representing the logic in terms of normalized position has two

advantages:

i) More intuitive understanding

ii) Adaptability

The electrohydraulic motor position at which the clutch is completely disengaged

varies between installations, and greatly depends on how well the system was bled. In

vehicle, for example, the motor position at complete disengagement is 1500°, while on the

bench test rig it was around 3300°. The Clutch Initialization Procedure, or Post Service

Procedure, is used to estimate the position of complete disengagement after servicing of

the clutch subsystem, or after the vehicle has been in use for some time.

When the CIP is activated, the logic ramps up the position request to the CCU, and

monitors the line pressure reported by the CCU. When a certain reference pressure is

reached, the logic uses the motor position at that point to estimate the motor position at

complete disengagement, and reports it as the ‘Max Disengaged Position’. The estimation

procedure is based on the position-pressure behaviour observed during bench testing of the

clutch.

It is seen from the position-pressure curves that depending on the air in the system,

the rise in pressure with increasing motor position is slow initially, as the air gets

compressed. After a certain pressure, the trend is linear and it becomes easier to estimate

the motor position at complete disengagement. This pressure is considered to be the

reference pressure. For the estimation algorithm, the reference pressure is taken to be 8 bar.

63
 From the standalone model developed for the clutch system, simulations were done

with differing air volumes in the lines. The curves generated were then fitted to a line from

8 bar to 18 bar of pressure, and data from bench testing was used to find the distance

between the point at 18 bar pressure and the disengagement point. Using this relationship,

a lookup table was created. This lookup table estimates the motor position at complete

disengagement once the position at 8 bar pressure is known. Table 11 shows a comparison

of actual disengagement positions and estimated disengagement position when the position

at the reference pressure was known.

Table 11: Comparison of CIP estimated disengagement positions

Comparison Position at 8 bar Estimated Actual

Disengagement Disengagement
Test Rig 1 3900 5196 5200
Test Rig 2 2150 3309 3300
Vehicle 550 1555 1500

The disengagement position estimated by the algorithm very closely matches the

actual disengagement for the test rigs, but is slightly higher for the vehicle. On the test rig,

it was very easy to feel the resistance on the clutch by spinning the transmission output

shaft. The exact point of disengagement could thus be recorded as the point where no

resistance was offered by the clutch. On the vehicle, however, complete disengagement is

taken to be the point where the wheels are able to spin freely on a lift with the transmission

locked in gear. This state has the clutch slipping at a low friction state. When the clutch is

completely disengaged, rattle is noticed as the clutch is ‘floating’. Even though the point

of complete disengagement on the vehicle is close to 1550, the disengaged state is

64
considered to be at 1500, so as to load the clutch up a bit to eliminate the rattle at a

disengaged state. The CIP was thus validated against three different setups with different

air amounts.

The estimation algorithm is also useful in knowing if complete disengagement of

the clutch is possible. With too much air in the system, a full stroke of the master piston

will not achieve complete disengagement of the clutch and the proper bleeding of the

system needs to be carried out. The CIP can also be carried out at certain intervals during

normal running, to adapt to changes in line air volume over time.

6.5 Bite Point Finding Procedure

The bite point, as defined earlier, is the clutch engagement which is enough to cause

vehicle movement by transmitting engine torque to the wheels (in 1st gear). A human driver

operating a car with a manual transmission for a while knows the bite point of that

particular vehicle and is able to release the clutch fast to that point during vehicle launch.

The vehicle launch clutch engagement strategy used in EcoCAR 3 vehicle follows a similar

approach, and it requires knowing the bite point of the clutch.

The procedure involves getting the vehicle on flat ground, and initializing the Bite

point Finding Procedure (BFP). The supervisory controller starts with a disengaged clutch,

transmission in first gear, and the engine running, and starts a ramped release of the clutch.

The clutch output speed is monitored and once movement is detected, the clutch is

disengaged and the clutch position at which movement was detected is recorded as the bite

point. The recorded bite point is then used as reference for clutch release during vehicle

65
launch. The procedure does not request torque from the engine, since the engine’s idle

controller usually generates enough torque to start pulling the car away.

The bite point estimate is around 1100 degrees motor position. The full

disengagement in-vehicle is at 1500 degrees, so the bite point is around 30% engagement,

which matches the prediction in Figure 32 in 5.4. The BFP process during a test on the

vehicle is plotted in Figure 38.

Figure 38: Execution of BFP on vehicle

6.6 Clutch Engagement Strategies

The hybrid architecture of the vehicle offers immense scope for use of the AMT. It

allows the clutch operation to be executed differently from a conventional IC engine

powertrain. The P1 motor in the EcoCAR vehicle, for example, makes it possible to better

66
control engine speeds during shift events, making speed matching possible and eliminating

the shift shock during clutch engagement. The P3 motor makes possible motor assisted

launches, thereby eliminating the need to slip the clutch to launch the vehicle. Certain

modes on the EcoCAR 3 vehicle, however, like the ‘Engine Only’ mode, make use of the

clutch the same way a conventional IC engine vehicle with a manual transmission does.

6.6.1 Clutch Engagement Types

Different types of clutch engagement options have been developed to suit the

engagement need of the vehicle operation mode. The vehicle mode operation section of the

supervisory control logic decides the engagement mode suitable for the vehicle mode, and

the Clutch Engagement Control section of the logic engages the clutch according to this

selection. Three clutch engagement options are available:

i) Direct engagement

ii) Counter based clutch engagement

iii) Assisted launch engagement

Direct Engagement -This is the simplest clutch engagement type and just involves

completely engaging the clutch without considering at component speeds or vehicle

speeds. This type is used during vehicle modes where the engine and transmission are not

needed, and the clutch is free to be engaged with the transmission in neutral. The mode

operation is responsible for ensuring that it is safe to engage the clutch, after verifying that

the transmission is in neutral. Modes where this engagement type is used are Startup, EV

Only, and Shutdown.

67
Counter Based Clutch Engagement – This is a comprehensive clutch engagement method

and caters to unassisted vehicle launch, gear shifting, and clutching in during a series –

parallel mode transition at speed. It is based on a counter to control clutch engagement.

The counter speed is based on the clutch output (transmission input) speed, and in essence,

involves a fast release of the clutch until the bite point, slow release through the bite region

until clutch lockup, and then fast release until complete engagement. The counter is used

during vehicle launch. This strategy also considers speed difference between the engine

speed and transmission input speed to provide for a gradual engagement.

Assisted Launch Engagement – Since the engine is not the only component producing

propulsive torque, the P3 REM can be used to get the vehicle moving, and the clutch is

then engaged when the transmission input shaft speed matches the engine speed.

6.6.2 Vehicle Launch

This section deals with the clutch engagement strategy employed during an ‘Engine

Only’ launch, where the IC engine is the only source of propulsive power and the launch

is similar to a conventional manual transmission vehicle. Different approaches were

considered to execute the launch. These approaches involved different control strategies to

modulate engine torque and regulate clutch position.

The ‘Counter Based Clutch Engagement’ uses the method that was introduced in

the previous section. The rationale behind developing this launch method was to make the

supervisory controller launch the vehicle the same way a human would.

This method controls release of the clutch based on the clutch output speed. When

the driver presses the accelerator pedal, the clutch release is started. The initial clutch

68
release is fast, and continues until bite point (set through BFP) is crossed. The logic then

switches over to the clutch speed based release table, which slows down the clutch release,

to make the clutch dwell near its bite point and engage gradually. Once the clutch output

speed is near engine idle speed, clutch release is again quickened and once clutch output

speed is close to engine output speed, the clutch is completely engaged fast, so that the

launch process is complete. The controller structure is shown in Figure 39. The counter

starts on receiving the enable signal, and is primarily controlled by the position and speed

based rates. The engine speed based compensator starts disengaging the clutch if the engine

speed drops below idle speed, and the speed difference based compensator slows down the

engagement if the speed difference between the engine and the clutch is high.

Figure 39: Counter based launch algorithm structure

The vehicle starts with the transmission in 1st gear, the engine idling, and the clutch

handshake completed, with the vehicle in a ready state. The inputs required by this logic

are listed in Table 12.

69
 Table 12: Important signals used by counter based launch method

Input Origin
Clutch Engagement Request Mode Operation
Engine Speed FPTCAN
Clutch Output Speed Input Conditioning

The counter is implemented as a resettable integrator, with the reset being the

Clutch Engagement Request, and the value to be integrated is the ‘rate’ that is provided by

a lookup table. The clutch output speed is estimated in the ‘Input Conditioning’ section of

the supervisory logic. This estimation makes used of the REM speed sent over CAN by the

REM inverter, and then uses the current gear to determine the transmission shaft input

speed. While the transmission does have a Transmission Output Shaft Speed (TOSS)

sensor, the REM speed is more reliable since it is a processed and filtered signal sent out

by the REM inverter, and is thus used to determine the clutch speed. The clutch output

speed is then used by the lookup table to determine the clutch engagement rate. An example

of this lookup table is shown in Figure 40.

70
 Figure 40: Clutch speed based release counter rate

The engagement logic also modulates the engine torque so that excessive

accelerator input does not cause high engine speeds during launch. The current strategy is

tailored for non-aggressive launches and limits the accelerator pedal input to the engine so

the engine speed stays below a set limit. The strategy does not exclusively control engine

torque, as some of the other approaches, but allows the driver to remain in control. The

clutch engagement strategy also considers engine speed, and overrides the counter to

disengage the clutch a bit if the engine speed starts falling below idle speeds, to avoid the

possibility of stalling the engine during the launch. An example of this strategy is shown

in Figure 41, with data for an ‘Engine Only’ launch on the vehicle.

71
 Figure 41: Vehicle launch with engine only

The clutch engagement trajectory is plotted from 100% (completely disengaged) to

0% (completely engaged). Vehicle movement is seen after clutch position crosses the bite

point. The clutch position trace exhibits the rapid release until bite point, dwell until lock

up, and rapid release until complete engagement. The clutch takes around 8 seconds to

completely engage.

This is a very slow vehicle launch, but the reason for the slow launch lies in the

very tall gearing ratio from engine to wheel even in 1st gear. A comparison of gear ratios

of the stock Camaro and OSU EcoCAR vehicle are shown in Table 13.

72
 Table 13: Comparison of gear ratios for OSU EcoCAR and stock Camaro

Vehicle Gear Ratio Differential Overall Ratio

Ratio
EcoCAR 3 2.95 2.77 8.17
2016 Camaro
4.40 3.27 14.39
1st Gear
2016 Camaro
2.59 3.27 8.47
2nd Gear

The overall ratio for the stock Camaro is 14.39 while that of the EcoCAR vehicle

is 8.17. This is taller even than the ratio of stock Camaro in 2nd gear. Launching the

EcoCAR vehicle in 1st gear is equivalent to launching the stock Camaro in 2nd gear. The

extremely tall gear ratio results in the slow vehicle launch.

To compare the controller based launch against a human driver launch, a third pedal

was added to the driver footwell as seen in Figure 42 and connected to the supervisory

controller. This third pedal was the same as the stock accelerator pedal, but the mounting

was done a bit differently to provide the feel of a clutch pedal. The pedal input was then

mapped inside the logic, with a 100% pedal press corresponding to 100% disengagement

in the clutch position request to the CCU. A switch was added inside the logic to be able

to switch between the driver clutch pedal and the logic requests.

The idea behind installing the temporary drive-by-wire clutch was to see how fast

a human launch was compared to the controller launch, compare the human clutch release

profile to that of the controller, and, if possible, use that trajectory as feedforward for

vehicle launch with a wraparound PI controller.

73
 Figure 42: Clutch pedal installation in vehicle

The clutch release profile for a human controlled launch are shown in Figure 43.

Similar vehicle launch times were seen when the driver was in complete control of both

the accelerator and clutch pedals, and the drivers perceived the launches to be difficult and

slow, and an aggressive launch usually resulted in the engine stalling.

The drivers first took a couple of dry runs to get used to the clutch pedal position

and feel. The initial launches were either very harsh or ended up stalling the engine. Once

the drivers got an idea of the bite point and the expected vehicle response, a clutch release

trajectory similar to the one produced by the controller was obtained. It is to be noted that

the clutch position trajectories are quite similar to the controller directed launch, with the

three distinctly defined regions.

The controller was also tested on the vehicle model in EcoSIM 3, and similar results

were obtained from the simulation, as seen in Figure 44.

74
Figure 43: Clutch release for human controlled launch

Figure 44: Vehicle launch in simulation

75
The results are the same as seen on vehicle, with the clutch taking around 8 seconds to

engage completely. The simulation was repeated for the stock Camaro ratios of 4.40 and

3.27 to obtain the results shown in Figure 45.

Figure 45: Simulation results for vehicle launch with stock Camaro gear ratios

With the stock Camaro ratios, the lockup phase is faster and clutch engagement

takes 5 seconds. If a more aggressive clutching strategy is used with the stock ratios, it is

possible to get a faster launch, as seen in Figure 46. The deeper ratios make it possible to

have a faster launch owing to higher wheel torque and faster engagement of the clutch.

When the aggressive strategy is simulated with the EcoCAR gear ratios, the engine stalls

in spite of the clutch starting to disengage. This is seen in Figure 47.

76
Figure 46: Simulation launch for stock ratios with aggressive controller

Figure 47: Simulated vehicle launch with EcoCAR gear ratios

77
6.6.3 Performance Launch

The vehicle has a ‘Performance Mode’ tailored for acceleration, and to retain the

feel of a performance vehicle. Launching the vehicle in performance mode uses the REM,

the engine, and the BAS to provide maximum wheel torque. The clutch engagement

strategy used during a performance launch is kept aggressive to get the engine torque to

the wheels as soon as possible.

The strategy uses the APP to determine the aggressiveness of the launch required.

Depending on how aggressive the driver wants the launch to be, the supervisory requests

varying torques from the REM, the engine, and the BAS, and also determines when to

engage the clutch. Below is the supervisory logic described for a WOT launch, which

would be the case when doing an acceleration test or a 0-60 mph run.

The supervisory logic demands maximum torque from all propulsive components.

At the start, just the REM provides full torque to get the car rolling, the clutch starts to

engage, and then the BAS and the engine are requested to provide full torque. An

aggressive clutch engagement strategy is possible during performance launch since the

combination of BAS torque and engine torque does not let the engine stall out during the

quick clutch engagement.

The clutch engagement during a performance launch can be seen in Figure 48.

78
 Figure 48: Clutch engagement during performance launch

Clutch engagement during a performance launch depends on the accelerator pedal

position and vehicle speed. Since component torque requests are linked directly to

accelerator pedal position, quickly releasing the clutch at lower APPs can result in stalling

the engine.

At lower APP, the launch is similar to an REM assisted launch, where the REM

provides torque at the start, and the clutch engages at a speed above engine idle speeds,

with the engine and BAS belting out torque post clutch engagement. BAS torque is

requested only above 60% APP.

79
 With the accelerator pedal floored, it indicates an actual performance launch

request. In this condition, the REM provides full torque at the start, the engine gets 10% of

the driver APP as its torque request as long as the clutch is disengaged. This allows the

engine to rev up a bit to avoid a stallout. At a speed of above 1 mph, the clutch is quickly

engaged, and the engine and BAS get full torque request when the clutch is more than 90%

engaged. The resultant clutch engagement is seen in Figure 48 and the engagement process

for a high APP and a low APP are shown in Figure 49.

Figure 49: Clutch engagement logic in performance mode

Clutch engagement for two extreme values of APP in performance mode are shown

in Figure 49, but the actual logic gradually transitions between these two extremes based

on APP. The clutch engagement speeds are defined based on the APP, and component

torques are requested based on the APP as well. A summary of the entire logic is given in

the next section.

80
 In Figure 48 during a full performance launch, it is seen that the engine speed

continues to be greater than the clutch speed i.e. the clutch continues to slip for about a

second even after the clutch is completely engaged. This is due to the fact that the combined

torque produced by the BAS and the engine is greater than the maximum transmissible

torque of the clutch at complete engagement. As was observed during torque

transmissibility test of the clutch, the clutch can transmit around 250 Nm of torque at

maximum engagement. While this is enough to handle the maximum torque produced by

the engine alone, the BAS at peak torque of 60 Nm causes an addition of an extra 160 Nm

of torque, leading to a maximum crankshaft output torque of ~ 320 Nm.

6.7 Vehicle Mode Based AMT management

Figure 50: Breakdown of AMT control tasks

The AMT logic is split into two sections in the code – the operational parameters

and high level requests are made in the ‘Mode Operation’ section that deals with vehicle

mode, and the actual operation is governed based on these requests in the ‘Actuators’
81
section of the supervisory logic. A breakdown of the actual tasks handled by each section

is shown in Figure 50.

The AMT management strategies, involving clutching, shifting, and component

torque management are described below for the major vehicle operation modes.

6.7.1 Vehicle Startup

The supervisory controller logic first goes into this mode when the vehicle is

started, either for the first time, or after a shutdown. This mode involves closing the right

relays, component startup handshakes, and transition into subsequent modes on successful

startup. The gear request is neutral and the clutch is commanded to engage after verifying

the transmission is in neutral. The direct engagement option is used to engage the clutch

since the vehicle is stationary.

6.7.2 EV Only Mode

EV only mode involves operating the vehicle using the REM alone. The engine and

BAS are woken up and enabled but not operational, and do not provide any torque. This

mode has the transmission in neutral and the clutch engaged.

6.7.3 Charge Sustaining Mode

The Charge Sustaining mode uses all torque producing components on the vehicle

and the torque is split to maintain the State Of Charge (SOC) of the battery pack. The

vehicle may operate either in series or parallel mode, depending on the torque request and

the SOC. In series mode, the clutch is disengaged to decouple the front powertrain from

the wheels, and the REM alone propels the wheels. The engine charges the battery pack

through the BAS, which acts like a generator. In parallel mode, the clutch is engaged and

82
the engine and REM both power the wheels, with the REM providing either propulsive

torque or performing regeneration, based on the SOC and torque request. The torque flows

during these two operations is shown in Figure 51.

Figure 51: Torque pathways in series and parallel operation

Gear selection in this mode depends on driver torque request. The most economic

gear is chosen based on vehicle speed and torque request. The counter based clutch

engagement type is chosen, and a gentler engagement table is selected. The counter based

engagement strategy is chosen since it also considers speed difference between the engine

and the clutch during engagement. This is essential since there might be a considerable

difference between these speeds when transitioning from series to parallel mode. The

counter does not have to be used for a vehicle launching, since this strategy allows a series

83
to parallel transition only when the clutch speed is above engine idle speed, and switches

to series mode if it falls below this speed. The clutch engagement in this mode is less

aggressive to minimize jerks, since the REM can provide wheel torque while the front

powertrain is decoupled.

6.7.4 Performance Mode

Performance mode has been implemented as a vehicle operation mode to retain the

feel of a performance vehicle. In this mode, all torque producing components are allowed

to supply peak torques to the wheels. This is a permanent parallel mode, with the front

drivetrain connected to the wheels at all times except launch and gearshifts. The

engagement strategy for the clutch has been dealt with in the previous section, and this

section looks at the high-level logic clutch engagements, gear changes, and torque requests.

Gear selection is based on a performance shift map, and a fast clutch release table

is used for quick clutch engagements. Transition to this mode can occur only at a standstill,

similar to getting the stock Camaro into a launch control mode. Transition into this mode

involves starting up the engine, disengaging the clutch and shifting the transmission into

1st gear.

Clutch engagement and disengagement requests are made based on accelerator

pedal position, Brake Pedal Position (BPP), and vehicle (clutch output) speed. The

following requirements were considered to determine the engagement / disengagement

speed thresholds:

i) Early engagement at high APP for faster launch

ii) Late engagement at low APP to avoid stalling

84
 iii) Early disengagement to avoid stalling under hard braking

iv) Disengagement speed should be lower than engagement speed.

Separate tables were made for engagement and disengagement speed thresholds, to

cater to the requirements listed above. Starting with the lowest speed that would allow the

clutch to engage without stalling the engine at 100% APP, the engagement threshold curve

was made for APP position. Starting with a disengagement speed that would not stall the

engine, the disengagement speed threshold curve was made for the APP position. The

curves were then slightly adjusted to maintain some gap between them to add hysteresis

and eliminate hunting. The resultant curves are shown in Figure 52.

Figure 52: APP based clutch speed thresholds

85
 At 100% APP, it was found that the clutch could be engaged at a clutch speed of

100 rpm. While this speed is very low compared to the engine idle speed, the engine and

BAS both produce propulsive torque at high APP, and the engine does not stall owing to

the combination of engine/BAS torque at the front and REM torque at the wheels, which

quickly raises clutch speed while the clutch engages.

At low APPs, the clutch disengagement speed threshold is set around 1000 -1200

RPM, which is slightly higher than the engine idle speed. The steepest threshold changes

occur above an APP of 80%, which is where the most aggressive launches are requested.

At APP below 60%, no BAS torque is requested, and hence the thresholds are near the

engine idle speed.

Different disengagement speed thresholds are required based on brake pedal

position. The disengagement thresholds have to be higher to allow for time to disengage

the clutch under heavy braking forces. If the same thresholds as Figure 52 are maintained,

there exists the possibility of the engine stalling out if the rapid deceleration does not allow

enough time for the clutch to disengage. Figure 53 shows the disengagement threshold

depending on the BPP.

The logic is set up so that in the event the BPP based table requests a disengagement

and the APP table requests an engagement (an unlikely occurrence, will occur only if the

accelerator pedal and the brake pedal are simultaneously pressed at particular values), the

logic will favour a disengagement. The fault detection part of the supervisory logic code

also sets the conditioned APP to zero if both pedals are pressed simultaneously.

86
 Figure 53: BPP based clutch speed thresholds

Braking below a BPP of 20% is subjectively considered to be light braking for the

EcoCAR 3 vehicle, and braking above 30% BPP is considered moderate to heavy. The

BPP based disengagement threshold speeds were set to ensure adequate time for

declutching the engine under very hard braking.

While there exist separate thresholds for APP and BPP, the maximum value

requested by each is taken into consideration when engaging or disengaging the clutch.

6.7.5 Series Mode

This mode operates the vehicle completely in series mode, and was designed to be

the limp home mode in case of a failure in the AMT system. This mode requests both

87
neutral gear and a disengaged clutch, to allow for the possibility of failure in either the

clutch or transmission subsystem.

6.7.6 Engine Only Mode

In this mode, the engine is the only source of propulsion, and this mode was

designed to be the limp home mode in case of failure in the HV torque producing

components. This mode operates in lower gears (1st /2nd/3rd) since the engine alone cannot

generate enough wheel torque at higher gears owing to the tall gearing ratios. The counter

based clutch engagement type is used and a gentler clutch engagement table is used. The

vehicle launch is done solely by the engine and the process mentioned in 2.3.1 is followed

in this mode. Clutch based speed thresholds similar to those in Figure 52 are used to request

clutch engagement / disengagement. The thresholds are again decided based on APP and

BPP, but the threshold tables differ from those shown in Figure 52 and Figure 53, since the

engine is the only torque producing component now and a conventional vehicle launch is

required by slipping the clutch.

There is no threshold table based on APP, since the brake is the only deciding factor

now that the engine is the only propulsive source. The clutching logic and the speed

thresholds are shown in Figure 54. The counter based clutch engagement strategy is used

for this mode since the engine is the only propulsive source, and the vehicle is now

equivalent to a conventional IC Engine vehicle. After vehicle launch, subsequent gearshifts

are handled by the counter based strategy as well, since it has a compensator to slowly

engage the clutch in case of a speed difference between the engine and the clutch.

88
 Figure 54: Clutching logic and speed thresholds for engine only mode

The BPP based speed thresholds are mentioned in terms of absolute vehicle speeds

since complete disengagement of the clutch is required when bringing the vehicle to a halt,

and the transmission is in 1st gear at these speeds. When the vehicle is at rest initially, the

clutch is disengaged and the transmission is in 1st gear. As the driver presses the accelerator

pedal, engagement is requested and the vehicle launch clutch engagement logic is

triggered. If the accelerator pedal is released while the vehicle is below the engagement

threshold, clutch disengagement is requested and the launch is aborted. If, however, the

vehicle is above the thresholds and the accelerator pedal is released, the clutch is still kept

engaged, as would be the case in a conventional manual transmission vehicle. If the

accelerator pedal is released and the speed falls below the disengagement threshold

(naturally or through application of brakes), the clutch is disengaged and the vehicle can

be stopped without stalling the engine.

89
6.7.7 Vehicle Shutdown

This mode involves the shutdown handshakes for all components and the power-

down procedure for the vehicle. The gear requested is neutral and clutch engagement is

requested after the transmission is confirmed to be in neutral. The clutch is engaged at

shutdown since that is the natural power-off state for the clutch. The transmission being in

neutral and the clutch being engaged are some of the required conditions for the ‘Shutdown

Complete’ signal being sent to mark a successful shutdown.

6.7.8 Parked Charging Mode

This mode is used for engine charging where the engine charges the ESS through

the BAS while the vehicle is in Park. In this mode, the transmission is kept in neutral and

the clutch is disengaged.

6.7.9 Miscellaneous Modes

Modes that do not involve any propulsive components, such as ‘Accessories Mode’

and ‘Charging mode’ have the transmission in neutral and the clutch engaged. In charging

mode, it does not matter since the vehicle relay is not closed and hence the SCU and CCU

do not have power.

90
 Chapter 7: Conclusions and Future Work

7.1 Accomplishments and Challenges

The team accomplished the task of automating the shifting process and the shifting

logic for its EcoCAR vehicle, which was one of the most important system level goals for

Year 3. The developed logic was robust and worked reliably throughout vehicle testing and

the Year 3 competition at General Motors’ Milford Proving Grounds in Milford, Michigan.

The extensive testing done by the team, both at a component level and the vehicle level,

ensured safe and reliable operation of the vehicle in Year 3 competition.

The team won the Year 3 competition held at Milford and Washington D.C., and

the work done for this thesis contributed to the following awards for individual events:

• dSPACE Embedded Success Award – 1st place

• Mathworks Modeling Award – 2nd place

One of the main challenges faced during the development of the AMT system

controls was the lack of inbuilt fault diagnosis on the SCU. The SCU is not capable of

telling the supervisory controller its status at a given time, and any faults it might have

encountered. This results in the supervisory controller having no control over the shift

itself; it can just command shifts. Another drawback is the lack of adaptability on the SCU-

it relies on good calibration by the user for mapping the gear positions, and does not go

through an initialization process. Moreover, since the supervisory controller cannot

91
command the SCU to move the motor to a certain position, it cannot execute a position

calibration procedure itself. This resulted in extra steps being added to the shifting logic

for safety and fault detection.

7.2 Future Work

The AMT system controller needs to be refined for drive quality and faster overall

shifting times. The team plans to use the BAS to perform speed matching during a gear

shift so that the clutch can be released fast while eliminating the shift shock and improving

the drive feel. The team also plans to change the gear ratios of the transmission and the

differential, making them deeper to have better launches, both in engine only and

performance mode. The Year 3 controls strategy uses a basic shift scheduling strategy, but

to maximize economy and performance, a better shift scheduling strategy needs to be

developed and optimized to reduce gear hunting and keep the engine at a point of high

efficiency. This requires validating the existing engine fuel map on the engine

dynamometer and recalibrating it, if necessary. To increase the fault detection capabilities

of the SCU, the team plans to develop an SCU of its own, incorporating sensors at different

locations and setting up the controller to be on CAN. These hardware and software

upgrades will ensure better utilization of the AMT system, a powerful tool in a hybrid

vehicle, while improving the drive quality and energy consumption of the vehicle.

92
 Bibliography

[1] “Alternative Fuels Data Center – Maps & Data” [Online]. Available:

https://www.afdc.energy.gov/data/

[2] Kluger, M. and Long, D., "An Overview of Current Automatic, Manual and

Continuously Variable Transmission Efficiencies and Their Projected Future

Improvements," SAE Technical Paper 1999-01-1259, 1999, doi:10.4271/1999-01-

1259.

[3] Greenbaum, J., Kluger, M., and Westmoreland, B., "Manual Transmission Efficiency

Trends and Characteristics," SAE Technical Paper 942274, 1994, doi:10.4271/942274.

[4] Wagner, G., "Application of Transmission Systems for Different Driveline

Configurations in Passenger Cars," SAE Technical Paper 2001-01-0882, 2001,

doi:10.4271/2001-01-0882.

[5] Wheals, J., Crewe, C., Ramsbottom, M., Rook, S. et al., "Automated Manual

Transmissions - A European Survey and Proposed Quality Shift Metrics," SAE

Technical Paper 2002-01-0929, 2002, doi:10.4271/2002-01-0929.

[6] Surampudi, B. and Wendel, G., "Control System Development for Retrofit Automated

Manual Transmissions," SAE Technical Paper 2009-28-0001, 2009, doi:10.4271/2009-

28-0001.

93
[7] Yacinthe, S., Khanna, A., Ward, J., Yatsko, M. et al., "Development of the Design of a

Plug-In Hybrid-Electric Vehicle for the EcoCAR 3 Competition," SAE Technical

Paper 2016-01-1257, 2016, doi:10.4271/2016-01-1257.

[8] Huang, H., Di, D., Chu, Y., and Guehmann, C., "Model-Based Optimization for an

AMT Clutch Control during the Vehicle Starting," SAE Int. J. Passeng. Cars –

Electron. Electr. Syst. 8(1):90-98, 2015, doi:10.4271/2015-01-0161.

[9] Tarasow, A., Bohn, C., Wachsmuth, G., Serway, R. et al., "Method for Identification

of the Kiss Point as well as Takeoff Point of a Hydraulically Actuated Friction Clutch,"

SAE Technical Paper 2012-01-0112, 2012, doi:10.4271/2012-01-0112.

[10] Haggag, S., "Sliding Mode Adaptive PID Control of an Automotive Clutch-By-Wire

Actuator," SAE Int. J. Passeng. Cars - Mech. Syst. 9(1):424-433, 2016,

doi:10.4271/2016-01-9106.

[11] Szadkowski, A. and Morford, R., "Clutch Engagement Simulation: Engagement

Without Throttle," SAE Technical Paper 920766, 1992, doi:10.4271/920766.

[12] L. Glielmo, L. Iannelli, V. Vacca and F. Vasca, "Gearshift control for automated

manual transmissions," in IEEE/ASME Transactions on Mechatronics, vol. 11, no. 1,

pp. 17-26, Feb. 2006.

[13] Haggag, S. and Ibrahim, F., "Dual Mode Control of an Automotive Clutch-By-Wire

System," SAE Technical Paper 2013-01-0484, 2013, doi:10.4271/2013-01-0484.

[14] Qin, G., Zhao, H., Lei, Y., and Ge, A., "A Strategy of On-board Fault Diagnosis of

Automated Mechanical Transmission," SAE Technical Paper 2000-01-1160, 2000,

doi:10.4271/2000-01-1160.

94
[15] Shimojo, K., Yasui, Y., Shimabukuro, E., and Iwaki, Y., "New Concept Sliding Mode

Control for AMT," SAE Technical Paper 2005-01-1594, 2005, doi:10.4271/2005-01-

1594.

[16] Breen, J. and Bower, G., "Clutchless Shifting of an Automated Manual Transmission

in a Hybrid Powertrain," SAE Technical Paper 2011-01-2194, 2011, doi:10.4271/2011-

01-2194.

[17] Yamasaki, M., Konno, H., Kuroiwa, H., and Ozaki, N., "Automated Manual

Transmission with Torque Assist Mechanism for Reducing Shift Shock," SAE

Technical Paper 2005-01-1783, 2005, doi:10.4271/2005-01-1783.

[18] Heath, R. and Child, A., "Zeroshift Automated Manual Transmission (AMT)," SAE

Technical Paper 2007-26-061, 2007, doi:10.4271/2007-26-061.

[19] Wang, W., Wang, Q., and Zeng, X., "Automated Manual Transmission Shift Strategy

for Parallel Hybrid Electric Vehicle," SAE Technical Paper 2009-01-0144, 2009,

doi:10.4271/2009-01-0144.

[20] Florêncio, D., Assis, E., and Amendola, C., "The Manual Transmission Automated -

Gearshift Quality Comparison to a Similar Manual System," SAE Technical Paper

2004-01-3363, 2004, doi:10.4271/2004-01-3363.

95