Dynamic Analysis and Control System Design of Automatic Transmissions

Downloaded from SAE International by University of Michigan, Monday, October 29, 2018
Dynamic Analysis and

Control System Design of
Automatic Transmissions
6293_Book.indb 1 1/11/13 3:51 PM

Other SAE books of interest:

Design Practices: Passenger Car Automatic
Transmissions, Fourth Edition
(Product Code: AE-29)
Innovations in Automotive Transmission Engineering

By Martin G. Gabriel
(Product Code: T-109)
Continuously Variable Transmission (CVT)

By Bruce D. Anderson and John R. Maten
(Product Code: PT-125)
For more information or to order a book,

contact SAE International at
400 Commonwealth Drive,
Warrendale, PA 15096-0001, USA;
phone 877-606-7323 (U.S. and Canada only)
or 724-776-4970 (outside U.S. and Canada);
fax 724-776-0790;
email CustomerService@sae.org;
website http://books.sae.org.
6293_Book.indb 2 1/11/13 3:51 PM

Dynamic Analysis and

Control System Design of
By Shushan Bai,
Joel Maguire, and Huei Peng
Warrendale, Pennsylvania
USA
Copyright © 2013 SAE International eISBN: 978-0-7680-7932-6
6293_Book.indb 3 1/11/13 3:51 PM

400 Commonwealth Drive

Warrendale, PA 15096-0001 USA
E-mail: CustomerService@sae.org
Phone: 877-606-7323 (inside USA and Canada)
724-776-4970 (outside USA)
Fax: 724-776-0790
Copyright © 2013 SAE International. All rights reserved.

No part of this publication may be reproduced, stored in a retrieval system, distributed, or transmitted, in any form
or by any means without the prior written permission of SAE International. For permission and licensing requests,
contact SAE Permissions, 400 Commonwealth Drive, Warrendale, PA 15096-0001 USA; email: copyright@sae.org;
phone: 724-772-4028; fax: 724-772-9765.
ISBN 978-0-7680-7604-2
SAE Order Number R-413
DOI 10.4271/R-413
Library of Congress Cataloging-in-Publication Data

Bai, Shushan.
Dynamic analysis and control system design of automatic transmissions / by Shushan Bai, Joel Maguire,
and Huei Peng.
pages cm
“SAE Order Number R-413.”
Includes bibliographical references.
ISBN 978-0-7680-7604-2
1. Automobiles—Transmission devices, Automatic. 2. Automobiles—Power trains. I. Maguire, Joel.
II. Peng, Huei. III. Title.
TL263.B35 2013
629.2′446—dc23 2012037412
Information contained in this work has been obtained by SAE International from sources believed to be reliable.
However, neither SAE International nor its authors guarantee the accuracy or completeness of any information
published herein and neither SAE International nor its authors shall be responsible for any errors, omissions,
or damages arising out of use of this information. This work is published with the understanding that SAE
International and its authors are supplying information, but are not attempting to render engineering or other
professional services. If such services are required, the assistance of an appropriate professional should be sought.
To purchase bulk quantities, please contact:

SAE Customer Service
Email: CustomerService@sae.org
Phone: 877-606-7323 (inside USA and Canada)
724-776-4970 (outside USA)
Fax: 724-776-0790
Visit the SAE International Bookstore at http://books.sae.org
6293_Book.indb 4 1/11/13 3:51 PM

Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Chapter 1 Automatic Transmissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Powertrain system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Why is a transmission necessary? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 First perspective: wheel power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Second perspective: engine operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Different types of automatic transmissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Chapter 2 Mechanics of Planetary Gear Automatic Transmissions . . . . . . . . . . 9

2.1 Torque converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.1 Description of torque converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.2 Dynamic model of engine and torque converter system . . . . . . . . . . . . . . . . 14
2.1.3 Backward calculation of torque converter variables . . . . . . . . . . . . . . . . . . . 18
2.2 Planetary gear trains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.1 Planetary gear set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.2 Planetary gear train . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.2.3 Static analysis of planetary gear trains: algebraic method . . . . . . . . . . . . . . . 26
2.2.4 Static analysis of planetary gear trains: lever analogy method . . . . . . . . . . . 29
2.2.5 Static analysis of planetary gear trains: matrix method . . . . . . . . . . . . . . . . . 33
2.2.6 Gear-shifting mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.2.7 Frictional clutches and their mathematical models . . . . . . . . . . . . . . . . . . . . 45
2.2.8 Dynamic equations of simple planetary gear sets . . . . . . . . . . . . . . . . . . . . . 47
2.2.9 Dynamic equations and simulation model of planetary gear trains . . . . . . 49
2.2.10 Generic dual-clutch model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.2.11 Matrix dual-clutch model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.2.12 Inertia balancing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.2.13 Six- and eight-speed planetary automatic transmissions . . . . . . . . . . . . . . . 65
6293_Book.indb 5 1/11/13 3:51 PM

Contents
Chapter 3 Control of Planetary Gear Automatic Transmissions . . . . . . . . . . . . 71

3.1 Electrohydraulic pressure control system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.1.1 Hydraulic pressure control system and its simulation models . . . . . . . . . . . . 73
3.1.2 Detailed model of PPC solenoids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.1.3 High-flow, direct-acting pressure control valves . . . . . . . . . . . . . . . . . . . . . . . 81
3.1.4 Pulse width modulated (PWM) solenoid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.1.5 Analytical study of hydraulic clutch control systems . . . . . . . . . . . . . . . . . . . . 84
3.2 Clutch-to-clutch gear-shift control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.2.1 Gear-shifting mechanics from a control perspective . . . . . . . . . . . . . . . . . . . 105
3.2.2 Hydraulic control system for clutch-to-clutch shift controls . . . . . . . . . . . . 110
3.2.3 Dynamic simulation model for studying clutch-to-clutch shift controls . . . 110
3.2.4 Control of power-on up-shifts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
3.2.5 Control of power-on down-shifts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
3.2.6 Hydraulic clutch control system design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
3.2.7 Clutch fill detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
3.2.8 Acceleration estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
3.2.9 Canceled shifts, transitional shifts, and double-transition shifts . . . . . . . . . 118
3.3 Electronic torque converter clutch control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.3.1 Hydraulic system for torque converter clutch control . . . . . . . . . . . . . . . . . . 121
3.3.2 Electronic control algorithm for torque converter clutch control . . . . . . . . 122
3.4 Dynamic analysis of torque converter clutch damper . . . . . . . . . . . . . . . . . . . . . . . . . . 126
3.5 Friction launch control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
3.6 Shift scheduling system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
3.6.1 Shift map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
3.6.2 Dynamic-programming-based shift map generation . . . . . . . . . . . . . . . . . . 137
3.6.3 Artificial intelligence-based shift scheduling system . . . . . . . . . . . . . . . . . . . 142
3.7 Integrated powertrain controls for driveability and fuel economy . . . . . . . . . . . . . . . . 150
3.7.1 Overall architecture of integrated powertrain control system . . . . . . . . . . . 150
3.7.2 Power-based gear selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
3.7.3 Constant-output-torque power-on up-shift control . . . . . . . . . . . . . . . . . . . 154
3.8 Centrifugal pendulum vibration absorber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
3.8.1 Basic concept and various designs for CPVA . . . . . . . . . . . . . . . . . . . . . . . . . 157
3.8.2 Equations of motion of CPVA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
3.8.3 Simulink model of CPVA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
3.8.4 Tuning of CPVA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
3.8.5 Path of pendulum motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
Chapter 4 Metal Pushing V-Belt Continuously Variable Transmissions . . . . . 167

4.1 Mechanics of metal pushing V-belt continuously variable transmissions . . . . . . . . . . 168
4.2 Controls of metal pushing V-belt continuously variable transmissions . . . . . . . . . . . 170
4.3 Comparison with other ratio control systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
4.4 Feed-forward/feedback control and its application to V-CVT ratio control . . . . . . . 177
vi
6293_Book.indb 6 1/11/13 3:51 PM

Contents
4.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177

4.4.2 Limitations of feedback controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
4.4.3 Feedback and feed-forward control systems . . . . . . . . . . . . . . . . . . . . . . . . . . 179
4.4.4 Design of feed-forward controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
4.4.5 Application: direct pulley pressure control of V-CVT . . . . . . . . . . . . . . . . . . 181
Chapter 5 Dynamics and Controls of Dual-Clutch Transmissions . . . . . . . . . . 185

5.1 Construction of dual-clutch transmissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
5.2 Synchronizer and its control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
5.3 Dual-clutch module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
5.4 Control algorithms for DCTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
About the Authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
vii
6293_Book.indb 7 1/11/13 3:51 PM

6293_Book.indb 8 1/11/13 3:51 PM

Introduction
Automatic transmissions for automobiles are not simply gearboxes, but have evolved into
complex integrated mechanical/electrical/hydraulic/electronic systems for achieving optimal
operation of vehicle powertrains. To achieve optimal design and control of modern automatic
transmissions, adequate dynamic analyses and control system designs are indispensable. This
book exclusively covers the topic of dynamic analysis and control system design of automatic
transmissions.
Why is dynamic analysis indispensable for achieving optimal design of automatic transmis-
sions? First, the gear shifting of automatic transmissions is a dynamic process, which involves
synchronized torque transfer from one clutch to another, smooth engine speed change,
engine torque management, and minimization of output torque disturbance. Dynamic
analysis is required to gain the necessary understanding of gear shifting mechanics, and
therefore to support the creation of an optimal design for gear shift control systems.
Hydraulic clutch control systems in automatic transmissions are highly dynamic systems.
Good dynamic and steady-state behaviors such as response time, stability, repeatability,
steady-state error, and robustness are the foundation for achieving premium gearshift quality.
The ratio of the continuously variable transmission (CVT) is controlled by hydromechanical
ratio feedback systems or electrohydraulic ratio feedback control systems. It is well known
that dynamic analysis is absolutely necessary for achieving adequate designs of hydraulic
control systems and feedback control systems.
In addition, an automatic transmission also provides isolation of engine torque pulsation
to reduce torsional vibrations of the powertrain and driveline. This is achieved through
torque converter clutch slip speed control and adequate design of the torque converter clutch
damper. In recent years, as a result of pursuing higher engine fuel efficiency, the magnitude
of the engine torque pulsation is increasing. To provide the required isolation for this higher
level of engine torque pulsations, adequate design of torque converter clutch dampers and
torque converter clutch control systems is necessary. This requires the use of dynamic-
analysis-based system design as well.
The electronic control algorithms for automatic transmissions include gearshift control
algorithms, gear schedule algorithms, torque converter clutch control algorithms, friction
launch control algorithms, and integrated powertrain control algorithms. Good designs
of such control algorithms must be based on a thorough understanding of the dynamic
characteristics of the systems as well as the use of the relevant dynamic analysis and design
methodologies.
ix
6293_Book.indb 9 1/11/13 3:51 PM

Introduction
While the basic working principle and the mechanical construction of automatic trans-
missions has not changed significantly, in recent years the tightening requirements for
performance, fuel economy, and drivability, as well as the increasing number of gears, have
made the design of transmission controls more challenging. The recent emergence of new
types of transmissions such as the Continuously Variable Transmission (CVT), Dual-Clutch
Transmission (DCT), and hybrid powertrain has presented added challenges. To respond to
such developments in automatic transmission technologies, this book will cover the broad
topics of dynamic analysis and control system design of automatic transmissions.
The book starts with the basic mechanics of automatic transmissions, and then covers in
detail topics of dynamics and controls of automatic transmissions. The topics covered include
gear-shifting mechanics and controls, dynamic modeling of planetary automatic transmis-
sions, design of hydraulic control systems, learning algorithms for achieving consistent
shift quality, torque converter clutch controls, the centrifugal pendulum vibration absorber,
friction launch controls, shift scheduling and integrated powertrain controls, CVT ratio
controls, and DCT controls, to name a few.
The book strives to provide a good balance between theory and practice. For the beginner,
the book will provide enough details to understand the basics of dynamics and controls
of automatic transmissions. For experienced engineers, this book will provide sufficient
theoretical discussions to help elevate the reader’s knowledge to a higher level. The book can
be used as a reference handbook for engineers as well as a teaching tool for classrooms. In the
sections that cover basic analytical skills, homework is provided for classroom usage.
6293_Book.indb 10 1/11/13 3:51 PM

Chapter 1
6293_Book.indb 1 1/11/13 3:51 PM

Chapter 1
An automatic transmission is a small component in automobiles, yet it plays an impor-

tant role in optimizing vehicle operations. This short, first chapter will provide a brief
description of the role of automatic transmissions and will serve as an introduction to
the book.
1.1 Powertrain system

The powertrain system in an automobile consists of an internal combustion engine and a
transmission, as shown in Fig. 1.1. The crankshaft of the engine is coupled to the input shaft of
the transmission through a launch unit. The launch unit could be a frictional clutch or a torque
converter. The output shaft of the transmission is connected to the vehicle wheels through
a final drive gear and a differential unit. A transmission has multiple speed ratios (torque
ratios and gear ratios). Speed ratio is defined as the ratio of transmission output speed to
input speed. Torque ratio is defined as the ratio of transmission output torque to input torque.
Therefore, speed ratio and torque ratio have reciprocal values. The torque ratio also equals
the ratio of input speed to output speed, and is referred to as the gear ratio. The gear ratio of
a transmission is selectable by a driver (in the case of a manual transmission) or by a pow-
ertrain controller (in the case of an automatic transmission) to achieve optimal operation of a
powertrain system. Typical transmission gear ratios are (for a six-speed transmission): 2.6 (1st
gear), 1.7 (2nd gear), 1.2 (3rd gear), 0.94 (4th gear), 0.77 (5th gear), and 0.62 (6th gear).
The first question one may ask is, “Why is a transmission necessary?” Let’s consider the answer
to this fundamental question before we start our discussions of dynamics and controls of auto-
matic transmissions.
1.2 Why is a transmission necessary?

To answer this question, let’s start by taking a look at the characteristics of internal combus-
tion engines. An internal combustion engine could be characterized by three graphs. The first
graph, Fig. 1.2, shows engine torque as a function of engine speed for each throttle opening.
The second graph, the engine power graph shown in Fig. 1.3, presents engine power as a
function of engine speed and throttle opening. The maximum engine power curve is the
100%-throttle-opening power curve labeled with “100% Th” on the graph. The third graph
is the BSFC (Brake Specific Fuel Consumption) graph shown in Fig. 1.4. BSFC is defined in
Fig. 1.1 Powertrain

system for automobiles.
6293_Book.indb 2 1/11/13 3:51 PM

Eq. 1.1. The BSFC graph shows that BSFC is a function of engine speed and engine torque. The
contour lines on the graph are the equal BSFC lines (iso-BSFC lines). The BSFC value of each
line is labeled on that line.
engine_fuel_rate
BSFC = (1.1)
engine_power
If an engine operates in the area where BSFC value is lower, higher fuel efficiency of opera-
tion could be obtained. Taking a careful note of the contour lines in Fig. 1.4, the minimum
BSFC line creates an island (210 g/(kW·h)). One of the tasks of a transmission is to maximize
Fig. 1.2 Engine torque graph.
Fig. 1.3 Engine power graph.
Fig. 1.4 Engine BSFC graph.
6293_Book.indb 3 1/11/13 3:51 PM

Chapter 1
the engine operations within this island while maintaining a balance with other drive-quality
considerations.
Now let’s answer the question of why a transmission is necessary. This question could be
answered from two perspectives. One is from the wheel power perspective. The other is from
the engine operation perspective.
1.2.1 First perspective: wheel power

Assume that a transmission only has one gear ratio. The maximum engine power curve then
can be placed on the vehicle speed coordinate by translating engine speed into vehicle speed
using gear ratio, final drive ratio, and wheel radius. The resulting line is called the propul-
sion power line and is shown in Fig. 1.5 as the solid line labeled 1st. The vehicle road load
power line, also shown in Fig. 1.5, is a broken line. It can be seen that with just one gear ratio,
the vehicle will run out of propulsion power at a low vehicle speed, and therefore the vehicle
cannot accelerate to a higher speed. If we have three additional higher gears to shift into before
the engine runs out of power, then three additional propulsion power lines can be added, as
shown in Fig. 1.6. The transition (shift) between gears occurs at the intersection points of pro-
pulsion lines. By doing so, the effective engine operating range is extended to the entire vehicle
speed range. If one would like to improve the first gear acceleration performance, and also
lower the engine speed at the top vehicle speed to improve fuel economy, two additional gears
can be added. The propulsion power lines of such powertrain systems are shown in Fig. 1.7.
Fig. 1.5 Propulsion power curve

with only one gear ratio.
Fig. 1.6 Propulsion power

curves of four gear ratios.
6293_Book.indb 4 1/11/13 3:51 PM

Fig. 1.7 Propulsion power

curves of six gear ratios.
1.2.2 Second perspective: engine operation

For a given vehicle speed, the transmission gear ratio determines engine speed. Therefore, the
transmission controls engine operating speed. Together with engine control algorithms, the
engine operating point (speed and torque) can be controlled. If we assume a driver demands
a certain level of propulsion power, then this driver’s power demand can be placed on the
engine BSFC graph as an equal power line, as shown in Fig. 1.8. Point A, which is the tangent
point between the equal power line and the “215” BSFC line, provides the best fuel efficiency
while delivering the driver’s power demand. If one selects the transmission gear ratio such that
the engine speed will be as close to point A as possible, and the engine controller constantly
ensures that the engine operates on the equal power line to achieve the driver’s power demand,
then the best fuel economy can be obtained.
A common explanation of the role of automatic transmissions is to transfer the engine power
to the vehicle wheels. However, from a control perspective, automatic transmissions are
actually engine speed control devices. Selecting the gear ratio of an automatic transmission
controls the engine speed. The engine speed and engine throttle opening together determine
the operating point of a powertrain. Therefore, the optimality (drive quality and fuel economy)
of a powertrain operation is largely determined by the design and control of the automatic
transmission.
There are a few additional needs for an automatic transmission as well. The typical automotive
engine has a minimum operating speed determined in part by combustion stability and firing
frequency limitation. This fact means that a starting device will have to be capable of allowing
Fig. 1.8 Constant power

line on BSFC graph.
6293_Book.indb 5 1/11/13 3:51 PM

Chapter 1
the engine to operate at some minimum speed while the transmission output speed (and
vehicle speed) is at or near zero. In addition, a reverse gear will also be required to propel the
vehicle in a reverse direction.
1.3 Different types of automatic transmissions

There are two basic types of automatic transmissions. One is the step gear transmission; the
other is the continuously variable transmission (CVT). Most step gear automatic transmis-
sions are planetary gear transmissions. Figure 1.9 is the cross-section view of a planetary gear
automatic transmission, which consists of a torque converter, a planetary gear train, and a gear
shift control system. While step gear automatic transmissions have discrete gear ratios, CVTs
can change input/output speed ratio continuously. Figure 1.10 is the cross-section view of a
metal pushing V-belt CVT. In a V-belt CVT, a hydraulic control system controls the operat-
ing radius of the belt around the driving and driven pulley to achieve the desired speed ratio.
A recent addition to the family of automatic transmissions, the Dual-Clutch Transmission
(DCT) shown in Fig. 1.11 is another type of step gear automatic transmission. DCTs make use
of parallel axis gears in place of planetary gears normally found in automatic transmissions.
However, the gear-changing mechanics and controls of both planetary and DCT transmissions
are fundamentally the same.
In the following chapters we will discuss the major dynamics and control issues of planetary
gear automatic transmissions, metal pushing V-belt continuously variable transmissions, and
dual-clutch transmissions.
Fig. 1.9 Cross-section view of a planetary gear automatic transmission.
6293_Book.indb 6 1/11/13 3:51 PM

Fig. 1.10 Cross-section view of a continuously variable transmission.
Fig. 1.11 Cross-section view of a dual-clutch transmission.
6293_Book.indb 7 1/11/13 3:51 PM

6293_Book.indb 8 1/11/13 3:51 PM

Chapter 2
Mechanics of Planetary Gear
6293_Book.indb 9 1/11/13 3:51 PM

Chapter 2
A planetary gear automatic transmission typically consists of a torque converter, a plan-

etary gear train, a set of clutches and bands, an electrohydraulic control system, and a
microprocessor controller with control software. Figure 2.1 is a conceptual diagram of
a typical planetary gear automatic transmission. Figure 2.2 is a cross-section view of a
planetary gear automatic transmission. The transmission depicted in Fig. 2.2 may not
seem cutting edge, but this well-known configuration contains the essential elements
found in today’s eight-, nine-, and ten-speed automatic transmissions.
In Fig. 2.1, P is the pump of the torque converter. The pump is connected to the engine
output shaft by a flex plate (not shown in Fig. 2.1). T is the turbine of the torque con-
verter. The turbine shaft is connected with the input shaft of the transmission. The
torque converter is a hydrodynamic power transfer device, and it functions as a vehicle
launch device and a torque multiplier. TCC is the Torque Converter Clutch. If the
TCC is engaged, the engine shaft and the transmission input shaft will be mechanically
locked together. R1, PC1, and S1 are the ring gear, pinion-carrier, and sun gear, respec-
tively, of the first planetary gear set. R2, PC2, and S2 are the ring gear, pinion-carrier,
and sun gear, respectively, of the second planetary gear set. The first and second plan-
etary gear sets are interconnected in a specific way to form a planetary gear train. The
other elements, such as C123 and CBR1, are clutches and bands. A desired gear ratio
is achieved by engaging the corresponding clutches and bands. Clutches, bands, and
the torque converter clutch are controlled by an electrohydraulic system (not shown in
Fig. 2.1).
Modern automatic transmissions are controlled by a microprocessor controller (not
shown in Fig. 2.1). The control algorithms programmed in the microprocessor usually
include a gear shift schedule algorithm, a gear shift control algorithm, a torque converter
clutch control algorithm, and diagnostics and fail-save control algorithms.
To build a foundation for the discussions of dynamics and control of automatic trans-
missions, this section introduces the major mechanical elements of planetary gear
automatic transmissions. These elements include torque converter, torque converter
clutch (TCC), planetary gear train, clutches, and bands. The description of those
elements in the following section is from the standpoint of control and dynamic
modeling of planetary gear automatic transmissions.
Fig. 2.1 Conceptual diagram of a four-speed planetary gear automatic transmission.
10
6293_Book.indb 10 1/11/13 3:51 PM

Mechanics of Planetary Gear Automatic Transmissions
Fig. 2.2 Cross-section view of a planetary gear automatic transmission.
2.1 Torque converter

2.1.1 Description of torque converter
Figure 2.3 is a cross-section view of a torque converter. Figure 2.4 is the conceptual diagram
of torque converters. The major components of a torque converter include the pump, turbine,
stator, and torque converter clutch. The vessel of a torque converter is filled with transmis-
sion fluid. The pump rotates with the engine shaft. The blades in the pump propel the oil onto
Fig. 2.3 Cross-section view of a torque converter.
11
6293_Book.indb 11 1/11/13 3:51 PM

Chapter 2
Fig. 2.4 Conceptual diagram

of a torque converter.
the blades in the turbine to transfer power from the engine to the transmission (see Fig. 2.5).
The fluid passages in the turbine redirect the flow of oil to a smaller diameter and into the
stator. The stator is grounded through a one-way clutch. Therefore, the stator can only rotate
in the direction of engine shaft rotation. When the turbine speed is low, the transmission fluid
coming from the turbine is redirected to the pump in the engine rotation direction by the
stator blades (see Fig. 2.6); this generates reaction torque on the stator in the engine rotation
direction. Because the summation of the engine torque, turbine torque, and stator torque is
zero, the resulting turbine torque is greater than the engine torque. This state of the torque
converter operation is called the torque multiplication state. When the turbine speed is high,
the stator will be freewheeling in the engine rotation direction, there will be no reaction torque
on the stator, and thus turbine torque will equal engine torque. This state of operation is called
the coupling state.
A torque converter is characterized by the following parameters:
Speed Ratio: n = Nt/Np = Turbine Speed/Pump Speed (2.1)
Torque Ratio: t = Tt/Tp = Turbine Torque/Pump Torque (2.2)
Input K-factor: Ki = Np/(Tp)1/2 (2.3)
Efficiency: E = (NtTt/NpTp)(100%) = n · t (100%) (2.4)
Fig. 2.5 The fluid flow

in a torque converter.
12
6293_Book.indb 12 1/11/13 3:51 PM

Fig. 2.6 The torque multiplication and

coupling states of torque converters.
The torque ratio t and input K-factor Ki are functions of the speed ratio n. A typical torque
ratio curve and input K-factor curve obtained from testing are shown in Fig. 2.7.
A torque converter is a power transfer device. There are four variables related to power: pump
torque, pump speed, turbine torque, and turbine speed. If two of four variables are known,
the other two can be obtained using the previous relationships. For example, if the pump and
turbine speeds are known, we can obtain the speed ratio from Eq. 2.1, and then using the input
K-factor curve and torque ratio curve shown in Fig. 2.7, we can obtain the value of the input
K-factor and torque ratio. With input K-factor and pump speed, the pump torque Tp can be
obtained from Eq. 2.3, and the turbine torque Tt can be obtained from Eq. 2.2.
When considering powertrain control optimizations, it becomes necessary to calculate pump
torque and pump speed for a given turbine torque and turbine speed, or to calculate turbine
torque and pump speed for a given turbine speed and pump torque. The method of those cal-
culations will be presented in Section 2.1.3.
In parallel with the hydrodynamic power transferring path just described, there is another
power transfer path: the torque converter clutch (see Fig. 2.4). The torque converter clutch is a
frictional clutch. When transmission fluid is fed into the TCC apply-port, the clutch is applied,
Fig. 2.7 Torque ratio, K-factor, and

efficiency as functions of speed ratio.
13
6293_Book.indb 13 1/11/13 3:51 PM

Chapter 2
and the engine shaft and turbine shaft are mechanically locked together. When fluid is fed into
the TCC release-port, the clutch is released. By controlling the pressure fed to the TCC apply-
port, the torque transferred through the TCC and the slip speed of torque control clutch (which
is the difference of pump speed and turbine speed) could be controlled to a desired level.
2.1.2 Dynamic model of engine and torque converter system

With the understanding of torque converter construction and operation in hand, this section
will introduce the first dynamic model in this text. The dynamic model of a system consists of
an engine and a torque converter. The model is used as an exercise to help the reader become
familiar with the process of developing dynamic models, as well as to establish a foundational
model that will be needed to build more complete models to simulate various transmission
control systems. This dynamic model also can help the reader to understand characteristics of
torque converters and torque converter clutches.
The system under consideration is shown in Fig. 2.8. In this system, the vehicle mass is
reflected onto the turbine shaft as the vehicle inertia Iv. Iv is combined with turbine inertia It.
The turbine and vehicle inertia, It + Iv, is driven by the torque converter turbine torque Tt and
torque converter clutch torque Ttcc. The pump inertia Ip is combined with engine inertia Ie.
The engine and pump inertia, Ie + Ip, is driven by the engine torque Te, and with pump torque
Tp and torque converter clutch torque Ttcc as the load torques. Notice that in Fig. 2.8, inertia
elements take torques as input and provide angular velocity to other elements, and the other
elements take angular velocity as input and provide torque to inertia elements. For example,
the torque converter takes engine speed we and turbine speed wt as inputs, and provides pump
torque Tp and turbine torque Tt to engine and pump inertia, and turbine and vehicle inertia,
respectively. Next, the dynamic equation of each element is introduced.
The first equation used to model the system, the equation of motion for the engine and pump
inertia, is shown in Eq. 2.5.
Te − Tp − Ttcc = (I e + I p )ω e + D pωe (2.5)
Here, we is the engine angular velocity, Ie is the engine inertia, Ip is the torque converter pump
inertia, Dp is the damping coefficient, Ttcc is the torque converter clutch torque, Tp is the torque
converter pump torque, and Te is the engine torque.
Fig. 2.8 Engine and torque converter system.
14
6293_Book.indb 14 1/11/13 3:51 PM

The engine torque Te can be modeled as a function of engine angular velocity we and engine
throttle opening, as shown in Eq. 2.6, and usually could be implemented with a two-dimen-
sional look-up table.
Te = f (ωe ,Th) (2.6)
The pump torque Tp can be calculated from the input K-factor Ki and engine angular velocity
we, as shown in Eq. 2.7.
ωe2
Tp = sign(ωe − ωt ) (2.7)
K i2
The input K-factor Ki can be obtained from test data as a function of torque converter speed
ratio n, as described in the previous section. Also notice that the pump speed equals the
engine speed.
The torque converter clutch torque Ttcc can be modeled using the hyperbolic-tangent function
shown in Eq. 2.8.
Ttcc = Ftcc K tcc tanh ((ωe − ωt ) / α ) (2.8)
Here, Ftcc is the torque converter clutch applying force; Ktcc is the torque converter clutch gain,
and a is a constant.
Equation 2.9 is the motion equation of turbine and vehicle inertia.
Tt + Ttcc = (I t + I v )ω t + Dt ωt (2.9)
Here, wt is the turbine angular velocity, It is the turbine inertia, Iv is the equivalent vehicle
inertia, Dt is the damping coefficient, and Tt is the torque converter turbine torque.
The turbine torque Tt can be obtained from Eq. 2.10.
Tt = Tp ⋅ t (2.10)
Here, the torque ratio t can be determined from the test data as the function of torque con-
verter speed ratio, as described in the previous section.
Using the presented dynamic equations, a Simulink model of the system can be built as
shown in Fig. 2.9. The submodels of the system are shown in Figs. 2.10–2.14. In the Simulink
model, the engine torque model, as shown in Eq. 2.6, is implemented using a look-up table.
The torque converter K-factor Ki and torque ratio r are also obtained from the corresponding
look-up tables as functions of torque converter speed ratio n.
A simulation result of the engine and torque converter system is shown in Fig. 2.15. In this
simulation, the TCC is released. It can be seen from the simulation that when the turbine
speed is low, the slip speed of the torque converter is high; therefore, speed ratio is low and
torque ratio is high, which result in higher turbine torque.
15
6293_Book.indb 15 1/11/13 3:51 PM

Chapter 2
Fig. 2.9 Simulink model of the engine and torque converter system (Main Model).
Fig. 2.10 Submodel of the Engine Torque Model.
Fig. 2.11 Submodel of the Engine Shaft Model.
16
6293_Book.indb 16 1/11/13 3:51 PM

Fig. 2.12 Submodel of the Torque Converter.
Fig. 2.13 Submodel of the TCC.
Fig. 2.14 Submodel of the engine and torque converter system (Turbine Shaft Model).
17
6293_Book.indb 17 1/11/13 3:51 PM

Chapter 2
Fig. 2.15 A simulation result of the engine and torque converter system.
2.1.3 Backward calculation of torque converter variables

A torque converter has four variables describing the state of its operation. Those variables are
turbine speed, turbine torque, pump speed, and pump torque. Knowing two of those variables,
the other two can be calculated using the input K-factor and torque ratio defined previously.
In a dynamic model of a powertrain system that includes a torque converter, the calculation
of torque converter variables is performed by calculating the pump torque and turbine torque
for the given pump speed and turbine speed, as shown in the previous section. Those calcula-
tions are defined in Eqs. 2.7 and 2.10. In some other applications such as integrated powertrain
controls, optimal gear selection, etc., it is necessary to perform backward calculations of
torque converter variables. Examples include calculating pump speed and pump torque for the
given turbine speed and turbine torque, or calculating pump speed and turbine torque for the
given pump (engine) power and turbine speed. This section will establish the method of these
calculations.
First, let’s consider the calculation of pump speed and pump torque for a given turbine
speed and turbine torque. The calculation method introduced here uses a Simulink model
to generate two look-up tables. One look-up table maps the given turbine speed and turbine
torque to pump speed. The other look-up table maps the given turbine speed and turbine
torque to pump torque. The Simulink model for generating such look-up tables is shown in
Figs. 2.16–2.18.
Figure 2.16 is the top-level Simulink model. The model generates two look-up tables: Kt_Tt_
Nt_to Ne and Kt_Tt_Nt_to_Te. Kt_Tt_Nt_to_Ne is the look-up table that maps the input
variable turbine speed and turbine torque to the output variable pump speed. Kt_Tt_Nt_to_Te
is the look-up table that maps the input variable turbine speed and turbine torque to the
output variable pump torque. Two look-up tables are saved in Workspace, and could be used
in any application to calculate pump speed and pump torque for the given turbine speed and
turbine torque. The two “From Workspace” blocks provide the input matrices Tt_matrix and
18
6293_Book.indb 18 1/11/13 3:51 PM

Fig. 2.16 Top-level Simulink model for generating look-up tables to map turbine torque and
turbine speed to pump speed and pump torque.
Fig. 2.17 Submodel for calculating pump speed and pump torque from given turbine speed and turbine torque.
Fig. 2.18 Submodel for calculating pump torque and turbine torque from given pump speed and torque converter speed ratio
using torque converter K-factor and torque ratio.
Nt_matrix to the computations. Tt_matrix and Nt_matrix have the following format, with
dimension of n rows and m columns.
 Nt_index 
 
 Nt_index 
Nt_matrix =  

 
 Nt_index 
Tt_matrix =  Tt_indexT Tt_indexT


Tt_indexT 
19
6293_Book.indb 19 1/11/13 3:51 PM

Chapter 2
Here, Nt_index is the row vector of turbine speed index for the look-up tables Kt_Tt_Nt_to Ne
and Kt_Tt_Nt_to_Te with size of m, and Tt_index is the row vector of turbine torque index for
the look-up tables with size of n.
The submodel with the name of “Generate look-up table for mapping turbine torque and
turbine speed to pump speed and pump torque” is shown in Fig. 2.17. This model uses closed-
loop control techniques to find the solution of pump speed and pump torque for the given
turbine speed and turbine torque. After running the model for a sufficient amount of time, the
variable labeled “error” will become very near zero, and then the values of Np and Tp are the
solution of pump speed and pump torque, respectively.
The submodel labeled “Torque Converter” in Fig. 2.17 is shown in Fig. 2.18. This model calcu
lates pump torque and turbine torque for the given pump speed and torque converter speed
ratio using torque converter input K-factor and torque ratio. The torque converter input K-factor
and torque ratio are provide by look-up tables as functions of torque converter speed ratio.
Look-up tables for calculating engine speed and turbine torque from given engine power and
turbine speed could be generated using the same method. The models for doing so are shown
in Figs. 2.19 and 2.20. Because the models are very similar to the model just described, expla-
nations of them are omitted.
Fig. 2.19 Top-level Simulink model for generating look-up tables to map engine power and turbine speed to engine speed and
turbine torque.
Fig. 2.20 Submodel for calculating engine speed and turbine torque from given turbine speed and engine power.
20
6293_Book.indb 20 1/11/13 3:51 PM

For advanced readers and MATLAB users, the following MATLAB program generates a
look-up table to map engine power and turbine speed to engine speed and turbine torque; and
a look-up table to map turbine power and turbine speed to engine speed and engine torque.
Kt_Pw_grid_for_Pw_Nt_to_Ne = [20:10000:800000 810000:100000:30000000]; % Nm*rpm

Kt_Nt_grid_for_Pw_Nt_to_Ne = [10:10:2500 2600:50:8000]; % rpm
SR_grid = [1:-0.001:0.001];
for k = 1:size(Kt_Nt_grid_for_Pw_Nt_to_Ne,2)
% engine power based

E_pw = Kt_Nt_grid_for_Pw_Nt_to_Ne(k)^3./(interp1(SR,K_factor,SR_grid).^2.*SR_grid.^3);
SR_v = interp1(E_pw, SR_grid, max(min(Kt_Pw_grid_for_Pw_Nt_to_Ne,max(E_pw)),min(E_pw)));
Te_v = (Kt_Nt_grid_for_Pw_Nt_to_Ne(k)./(interp1(SR,K_factor,max(min(SR_v,max(SR)),min(SR))).*SR_v)).^2;
Kt_ePw_Nt_to_Ne(:,k)= Kt_Nt_grid_for_Pw_Nt_to_Ne(k)./SR_v;
Kt_ePw_Nt_to_Tt(:,k)= Te_v.*interp1(SR,TR,max(min(SR_v,max(SR)),min(SR)));
% turbine power based

T_pw = interp1(SR,TR,SR_grid).*Kt_Nt_grid_for_Pw_Nt_to_Ne(k)^3./(interp1(SR,K_factor,SR_grid).*SR_grid).^2;
SR_v_1 = interp1(T_pw, SR_grid, max(min(Kt_Pw_grid_for_Pw_Nt_to_Ne,max(T_pw)),min(T_pw)));
Kt_tPw_Nt_to_Te(:,k) = …
(Kt_Nt_grid_for_Pw_Nt_to_Ne(k)./(interp1(SR,K_factor,max(min(SR_v_1,max(SR)),min(SR))).*SR_v_1)).^2;
Kt_tPw_Nt_to_Ne(:,k) = Kt_Nt_grid_for_Pw_Nt_to_Ne(k)./SR_v_1;
end
Homework for Section 2.1

1. Implement the dynamic model of the engine and torque converter system
in Simulink. Run simulations at several different throttle levels. Discuss what
characteristics you would like to have for mating the torque converter to a high-
speed engine.
2.2 Planetary gear trains

Planetary gear trains are the central part of planetary gear automatic transmissions. A plan-
etary gear train is a group of interconnected planetary gear sets. The simple planetary gear sets
will be introduced first, and then the construction of planetary gear trains and planetary gear
automatic transmissions will be presented. Building on the understanding of the construc-
tions of planetary trains, the static analysis methods of planetary gear trains will be introduced.
The introduction of the static analysis methods includes the algebraic method, lever analogy
method, and matrix method. As these fundamentals are established, our attention will be
extended to the gear-shifting mechanics. First, we will use static analysis methods to establish
basic understandings of gear-shifting mechanics. Then, the major focus of this section, the
dynamic modeling and analysis of planetary gear trains, will be presented.
2.2.1 Planetary gear set

A simple planetary gear set is shown in Fig. 2.21. There are four components in a simple
planetary gear set: sun gear, ring gear, planet gears, and planet carrier. The planet carrier, sun
gear, and ring gear are concentric with the main axis. There are typically three planet gears
(although it is possible to construct a simple planetary system with only two planet gears, and
the modern planetary can have as many as six planet gears). The planet gears are mounted on
21
6293_Book.indb 21 1/11/13 3:51 PM

Chapter 2
Fig. 2.21 Simple planetary gear set.
the planet carrier via planet pins that are fixed to the planet carrier, and can rotate about the
planet pin axis as well as rotate with the carrier about the main axis.
A simple planetary gear set can be represented by a stick diagram as shown in Fig. 2.22, or by a
lever diagram as shown in Fig. 2.23 [2-1]. In those figures, RG, SG, and C represent ring gear,
sun gear, and planet carrier, respectively. The length of the lever is proportional to the number
of teeth of the sun gear and ring gear, S and R, as shown in Fig. 2.23.
The speeds of the sun gear, ring gear, and planet carrier are not independent, and are related by
Eq. 2.11. Here, S and R are the radii (or number of teeth) of the sun gear and ring gear, respec-
tively, and ws, wr, and wc are speed of the sun gear, ring gear, and carrier, respectively. Equation
2.11 is called the speed constraint equation of a simple planetary gear set.
ωs S + ωr R = ωc (R + S ) (2.11)
Fig. 2.22 Stick diagram of

a simple planetary gear set.
Fig. 2.23 Lever diagram of

22
6293_Book.indb 22 1/11/13 3:51 PM

The steady-state torque balance equations of a simple planetary gear set are given in Eqs. 2.12
and 2.13.
Ts + Tr − Tc = 0 (2.12)
Tr S = Ts R (2.13)
Here, Ts, Tr, and Tc are the torque exerted on the sun gear, ring gear, and carrier, respectively.
2.2.2 Planetary gear train

In a planetary gear automatic transmission, multiple planetary gear sets are interconnected
to achieve the desired number of gears and gear ratios. A group of interconnected planetary
gear sets is called a planetary gear train. Figure 2.24 shows the stick diagram of a planetary
gear train consisting of two parallel-connected planetary gear sets. Here, the first ring gear
RG1 is connected with the second planet carrier C2, and the first carrier C1 is connected
with the second ring gear RG2. The cross-section of this parallel-connected planetary gear
train is shown in Fig. 2.25. Figure 2.26 is the lever diagram of this planetary gear train. The
lever diagram in Fig. 2.26 can be further simplified as shown in Fig. 2.27. Note that the length
of each section of the simplified lever indicated in Fig. 2.27 is proportionally rescaled. The
method of rescaling and the details of the lever diagram analogy of planetary gear trains will
be discussed in the following sections.
The speed constraint equations of the planetary gear train presented in Figs. 2.24–2.27 are
shown in Eqs. 2.14–2.17. Equations 2.14 and 2.15 are the speed constraint equations of the two
Fig. 2.24 Stick diagram

of a planetary gear train.
Fig. 2.25 Cross-section

23
6293_Book.indb 23 1/11/13 3:51 PM

Chapter 2
Fig. 2.26 Lever diagram

Fig. 2.27 Simplfied lever diagram

planetary gear sets. Equations 2.16 and 2.17 define the connections of the two planetary gear
sets.
ωs1S1 + ωr 1R1 = ωc 1 (R1 + S1 ) (2.14)
ωs 2S2 + ωr 2 R2 = ωc 2 (R2 + S2 ) (2.15)
ωr 1 = ωc 2 (2.16)
ωr 2 = ωc 1 (2.17)
In the following discussions, we define the elements of the sun gear, ring gear, and carrier
as nodes. By connecting certain nodes to the input shaft, output shaft, or ground (meaning
connected to the transmission case, so the node cannot rotate), an input-output gear ratio
can be achieved. For the previous planetary gear train, the first gear is accomplished by con-
necting node SG1 to the input, connecting node C1&RG2 to the output, and grounding node
RG1&C2. The expression of the first gear ratio can be obtained from Eqs. 2.14–2.17 by letting
wr1 = wc2 = 0, ws1 = win, and wr2 = wc1 = wout. Here, win and wout are the input and output speeds,
respectively. Then, as shown in Eq. 2.18, we have
ωin R1 + S1
= (2.18)
ωout S1
The gear ratios of other gears can be obtained in the same manner. Table 2.1 summarizes the
expression of all gear ratios of the planetary gear train shown in Fig. 2.24.
As shown in Table 2.1, to achieve a desired gear, certain nodes need to be connected to the
input, certain nodes need to be connected to the output, and certain nodes need to be con-
nected to the ground. In automatic transmissions, such connections are accomplished by a set
of frictional clutches. Figure 2.28 shows how this is realized. In Fig. 2.28, C123, C24, C34, C1R,
24
6293_Book.indb 24 1/11/13 3:51 PM

Table 2.1 Gear Ratios of the Planetary Gear Train of Fig. 2.24
Gear Connections Gear Ratio Expression
1st SG1 input, ws1 = win ω in R1 + S1

=
C1&RG2 output, and wr2 = wc1 = wout ω out S1
RG1&C2 ground, wr1 = wc2 = 0
2nd SG1 input, ws1 = win ω in R1S2 + S1R2 + S1S2

=
C1&RG2 output, and wr2 = wc1 = wout ω out S1S2 + S1R2
SG2 ground, ws2 = 0
3rd SG1 input, ws1 = win ω in

=1
RG1&C2 input, wr1 = wc2 = win ω out
C1&RG2 output, and wr2 = wc1 = wout
4th SG2 ground, ws2 = 0 ω in R2

=
RG1&C2 input, wr1 = wc2 = win ω out R2 + S2
Reverse SG2 input, ws2 = win ω in R

=− 2
RG1&C2 ground, wr1 = wc2 = 0 ω out S2
Fig. 2.28 Stick diagram of a four-speed planetary gear automatic transmission.
and CR are frictional clutches. Figure 2.29 presents the lever diagram of the system. Table 2.2
shows the clutch states for each gear; X means the clutch is engaged.
As mentioned previously, the connections of nodes to input, output, or ground are accom-
plished by frictional clutches. The cross-section of a frictional clutch is shown in Fig. 2.30. The
frictional plate set A is splined on the shaft A. The friction plate set B is splined on the shaft B.
A hydraulic piston is connected to shaft A. As the cylinder of the hydraulic piston is pressur-
ized, the piston pushes the friction plate sets A and B together, and connects shaft A with shaft
B. The torque capacity of the clutch, which is the maximum torque the clutch can transfer, is
determined by the applied hydraulic force, clutch radius, number of plates, and coefficient of
friction. The torque exerted on the shafts determines the actual torque carried by the clutch. If
the actual clutch torque exceeds the torque capacity, the clutch will slip.
25
6293_Book.indb 25 1/11/13 3:51 PM

Chapter 2
Fig. 2.29 Lever diagram of a four-speed planetary gear automatic transmission.
Table 2.2 Clutch State Table

Gear C123 C1R C24 C34 CR
1st gear X X
2nd gear X X
3rd gear X X
4th gear X X
Reverse gear X X
2.2.3 Static analysis of planetary gear trains:

algebraic method
Static analysis of a planetary gear train includes calculations of speed ratio and torque ratio.
Three methods can be used: the algebraic equation method, lever analogy method, and matrix
method. This section discusses the algebraic equation method. Section 2.2.4 discusses the lever
analogy method, and Section 2.2.5 discusses the matrix method.
The speed ratio is defined as the ratio of the speeds of two nodes in a planetary gear train.
The input speed ratio, which is the ratio of the input speed node vs. the speed of another node,
is used the most in planetary gear train analysis. The speed ratios can be calculated from the
speed constraint equations of planetary gear sets and connection equations. Let’s consider the
transmission system in Fig. 2.28 again. The speed constraint equations for the two planetary
gear sets are given in Eqs. 2.19 and 2.20. Two permanent connection equations are given in
Eqs. 2.21 and 2.22. Other connection equations are gear-state dependent.
ωs1S1 + ωr 1R1 = ωc 1 (R1 + S1 ) (2.19)
ωs 2S2 + ωr 2 R2 = ωc 2 (R2 + S2 ) (2.20)
ωr 1 = ωc 2 (2.21)
26
6293_Book.indb 26 1/11/13 3:51 PM

ωr 2 = ωc 1 (2.22)
For the first gear, because node SG1 is connected to the input through clutch C123, node
RG1&C2 is grounded through clutch C1R, and node C1&RG2 is the output, the additional
connection equations are given in Eqs. 2.23–2.25.
ωin = ωs1 (2.23)
ωout = ωr 2 = ωc 1 (2.24)
ωr 1 = ωc 2 = 0 (2.25)
Substituting the connection equations into the planetary gear speed constraint equations, Eqs.
2.19 and 2.20, Eqs. 2.26 and 2.27 are obtained.
ωinS1 + 0 ⋅ R1 = ωout (R1 + S1 ) (2.26)
ωs 2S2 + ωout R2 = 0 ⋅ (R2 + S2 ) (2.27)
From Eqs. 2.26 and 2.27, the speed ratio Eqs. 2.28 and 2.29 are obtained.
ωin R1 + S1
= (2.28)
ωout S1
ωin (R + S )S
=− 1 1 2 (2.29)
ωs 2 S1R2
Equations 2.23, 2.25, 2.28, and 2.29 define the input speed ratio of all elements when the plan-
etary gear train is in first gear. Speed ratios of the other gears can be obtained in the same way.
Fig. 2.30 Cross-section of a frictional clutch.
27
6293_Book.indb 27 1/11/13 3:52 PM

Chapter 2
The torque ratio is defined as the ratio of the torques between two nodes in a planetary
gear train. Two types of torque ratios are of most interest. One is the input torque ratio of a
clutch, which is defined as the torque ratio of a clutch node vs. the input node. Another one
is the output torque ratio of a clutch, which is defined as the torque ratio of the output node
vs. a clutch node. Torque ratios can be obtained from the torque equations of planetary gear
sets and torque relationship equations of a planetary gear train. Take the second gear as an
example. After eliminating non-engaged clutches, the free-body diagram for the second gear is
obtained as shown in Fig. 2.31.
Using the torque equations of a planetary gear set given in equations 2.12 and 2.13, and the
torque relationship equations Ts1 = Tc123, Tr1 = Tc2, Ts2 = Tc24, and Tr2 = Tc1 – Tout, the torque
equations for the second gear are obtained from the free-body diagram, as shown in
Eqs. 2.30–2.34.
Tc 2 + Tc 123 = Tc 1 (2.30)
Tc 2S1 = Tc 123R1 (2.31)
Tc 1 + Tc 24 = Tc 2 + Tout (2.32)
(Tc 1 − Tout )S2 = Tc 24 R2 (2.33)
Tc 123 = Tin (2.34)
From these torque equations, and some algebraic manipulations, the input torque ratios of
clutches are obtained, as shown in Eqs. 2.35–2.38. Equation 2.39 is the output/input torque
ratio. The clutch output torque ratios are shown in Eqs. 2.40–2.42.
Tc 123
=1 (2.35)
Tin
Tc 24 R1S2
= (2.36)
Tin S1 (R2 + S2 )
Tc 1 R1 + S1
= (2.37)
Tin S1
Fig. 2.31 Free-body

diagram of second gear.
28
6293_Book.indb 28 1/11/13 3:52 PM

Tc 2 R1
= (2.38)
Tin S1
Tout S1 (R2 + S2 ) + R1S2

= (2.39)
Tin S1 (R2 + S2 )
Tout S1 (R2 + S2 ) + R1S2

= (2.40)
Tc 123 S1 (R2 + S2 )
Tout S1 (R2 + S2 ) + R1S2 (2.41)

= ⋅
Tc 24 R1S2
Tout S1 (R2 + S2 ) + R1S2 (2.42)

=
Tc 1 (R1 + S1 )(R2 + S2 )
2.2.4 Static analysis of planetary gear trains: lever

analogy method
The algebraic equation method is straightforward, but it requires laborious algebraic works.
The lever analogy method [2-1] reduces algebraic works, and provides a visualized representa-
tion of speed and torque ratios.
The lever analogy method is based on the idea of using a lever to represent a planetary
gear set. This method is well documented in the SAE International Paper No. 810102, “The
Lever Analogy: A New Tool in Transmission Analysis,” by Howard L. Benford and
Maurice B. Leising.
A lever shown in Fig. 2.32 is adapted to analogize a simple planetary gear set. The length of
each section of the lever corresponds to the radius (or number of teeth) of sun and ring gears,
as labeled on the lever. Three nodes labeled SG, RG, and C represent sun gear, ring gear, and
carrier, respectively. As explained in the paper, the reason why a lever can be used to analogize
a planetary gear set is that the speed constraint equation and torque equations of a planetary
gear set are preserved in the lever analogy world. Figure 2.33 is the lever diagram of simple
planetary gear set for speed ratio calculation. The vertical lines from the zero speed line rep-
resent sun, ring, and carrier speed. It can be easily proven using the similar strangler theorem
that the speed equation, Eq. 2.11, is preserved in the lever diagram. Figure 2.34 is the lever
diagram of a simple planetary gear set for torque ratio calculation. Torque Equation 2.12 can
be obtained by taking the zero summation of all torques in the vertical direction. Torque
Equation 2.13 can be obtained by applying the principle of balance of force moment about
node C. Here, torque is treated as force in the lever method.
Fig. 2.32 Lever analogy of

29
6293_Book.indb 29 1/11/13 3:52 PM

Chapter 2

a simple planetary gear set
for speed ratio calculation.

a simple planetary gear set
for torque ratio calculation.
A planetary gear train can be analogized by interconnecting simple levers, as shown in Fig.
2.35. Since this two-lever system is connected in parallel, it can be simplified further into one
lever, as shown in Fig. 2.36. In this figure, the fixed links between C1-RG2 and RG1-C2 are the
parallel connections.
In the simplified lever diagram, the length of the section from the node RG1&C2 to the node
SG2 is rescaled such that the ratio relationship in Eq. 2.43 is satisfied. Therefore, the simplified
lever diagram preserves all ratio relationships of the original two-lever system.
length_from_SG2_to_RG1&C2 R2
= (2.43)
length_from_RG1 & C2_to_C1&RG2 S2
The speed ratios can be obtained easily from lever diagrams. Taking the first gear as an
example, the lever position of the first gear is shown in Fig. 2.37. In Fig. 2.37, the horizontal
line represents the zero speed line. Because the node RG1&C2 is grounded in the first gear,
the first gear line rotates about the RG1&C2 node. The length of the vertical line from each
node to the first gear line represents the speed of each node in the first gear. Using the similar
Fig. 2.35 Lever diagram of the

four-speed planetary gear train.
Fig. 2.36 Simplified lever diagram of

the four-speed planetary gear train.
30
6293_Book.indb 30 1/11/13 3:52 PM

triangle property, the speed ratios shown in Eqs. 2.28 and 2.29 can be easily obtained, as dem-
onstrated in the following equation.
ω in length_of_input_speed_line length_from_SG1_to_RG1&C2 R1 + S1
= = =
ω out length_of_output_sped_line length_from_C1&RG2_to_RG1&C2 S1
The previous expression proves Eq. 2.28. Equation 2.29 can be proven in the same manner.
The lever position lines of other gears are shown in Figs. 2.38–2.41.
Fig. 2.37 Lever position

line of first gear.

line of second gear.

line of third gear.

line of fourth gear.
31
6293_Book.indb 31 1/11/13 3:52 PM

Chapter 2

line of reverse gear.

a four-speed planetary gear
automatic transmission.
At this point, clutches can be included in the simplified lever diagram. As the result, the
complete lever diagram of the four-speed planetary gear automatic transmission is obtained, as
shown in Fig. 2.42.
The torque ratios can also be easily obtained from the lever diagrams. Let’s take the first gear
as an example. Torques are added on the lever as shown in Fig. 2.43. In the lever analogy of
planetary gear trains, torque is analogically treated as force.
The torque ratios can be obtained by applying the principle of balance of moment. Consider-
ing the force moment balance about the node RG1&C2, then Eq. 2.44 can be obtained. This
equation defines the output/input torque ratio. Similarly, the force moment balance equation
about node C1&RC2 is obtained as shown in Eq. 2.45, which defines the C1R/input torque ratio.
Tin (R1 + S1) = Tout S1 (2.44)
Tin R1 = TC 1R S1 (2.45)
Torque ratios of other gears can be obtained in the same manner.
Fig. 2.43 First gear torques

on the lever diagram.
32
6293_Book.indb 32 1/11/13 3:52 PM

2.2.5 Static analysis of planetary gear trains:

matrix method
The planetary gear train shown in Fig. 2.28 will be used once more to illustrate the matrix
method [2-2, 2-3]. First, let’s consider how to use the matrix method to calculate speed ratios.
The two speed constraint equations of two planetary gear sets can be rewritten into the form
shown in Eqs. 2.46 and 2.47. Two permanent connection equations can be rewritten into the
form shown in Eqs. 2.48 and 2.49.
ωs1S1 + ωr 1R1 − ωc 1 (R1 + S1 ) = 0 (2.46)
ωs 2S2 + ωr 2 R2 − ωc 2 (R2 + S2 ) = 0 (2.47)
ωr 1 − ωc 2 = 0 (2.48)
ωr 2 − ωc 1 = 0 (2.49)
Let’s also consider the case of first gear. Because the first ring gear RG1 is grounded in the first
gear, we have the grounding equation, Eq. 2.50. Assume that we are interested in the speed
ratio from all nodes to the input node, which is the first sun gear. Therefore, we add Eq. 2.51.
ωr 1 = 0 (2.50)
ωin = ωs1 (2.51)
The previous six equations can be put into a matrix format, as shown in Eq. 2.52. In the matrix
M, the first two rows are from two speed constraint equations of planetary gear sets. The third
and forth rows are the connection equations. There is one element with value of 1, and one
element with value of –1 in each connection row, and the rest of the elements are zeros. The
location of 1 and –1 correspond to the connected nodes. The fifth row has one element with
value of 1, and the rests are zeros. The location of 1 corresponds to the grounding node. The
location of 1 in the sixth row corresponds to the input node. The matrix B has 1 at the last
element, and the rest are zeros, because the last row of matrix M defines the location of the
input node.
Mω = Bωin (2.52)
Here,
 S1 R1 −(S1 + R1 ) 0 0 0 
 
 0 0 0 S2 R2 −(S2 + R2 ) 
 
M = 0 −1 0 0 0 1 
 0 0 1 0 −1 0 
 0 1 0 0 0 0 
 
 1 0 0 0 0 0 
33
6293_Book.indb 33 1/11/13 3:52 PM

Chapter 2
 ωs 1 
 
 ωr 1 
 
ωc 1 
ω = 

ωs 2 
 
 ωr 2 
 
 ωc 2 
 0 
 
 0 
 0 
B = 
 0 
 0 
 1 
The solution of Eq. 2.52 is shown in Eq. 2.53. The elements of vector r are the speed ratios
from each node to the input node S1.
ω = M−1Bωs1 = rωin (2.53a)
Here,
 rs1/s1 
 
 rr 1/s1 
 
rc 1/s1 
r = 
rs 2/s1 
 
 rr 2/s1 
 
 rc 2/s1 
We can rewrite the result as shown in Equation 2.53b.
ωs1   rs1/s1 

  
ωr 1   rr 1/s1 
ωc 1   
rc 1/s1 
 = ωin (2.53b)
ωs 2   rs 2/s1 
ω   
rr 2/s1 
 r2  
ωc 2   rc 2/s1 

34
6293_Book.indb 34 1/11/13 3:52 PM

Notice that in the matrix M, the first two rows are the speed constraint equations of two plan-
etary gear sets. The third and fourth rows are the speed constraint equations of two connected
nodes. Those rows define the connections between nodes, and have one element with value
of 1, one element with value of –1, and the rest of the elements with value of 0. The fifth row
is the speed constraint equation of the grounding node. This row defines the grounded node,
and has one element with value of 1, and the rest of the elements with value of 0. The last row
defines the input node; therefore, it also requires the last element of vector B, which is 1. From
the previous observations, one can directly write the matrix M and B for any planetary gear
train system without taking note of the speed constraint equations and connection equations
first. For the matrix equation to be solvable, the total number of speed constraint equations,
connection equations, and grounding node equations has to be n – 1. Here, n is the length of
the speed vector ω .
Next, let’s consider how to use the matrix method to calculate torque ratios. To do this, let’s
start from the fundamental step: the free-body diagram. Based on our understanding of
the result obtained from this fundamental step, we will arrive at a simpler and more direct
matrix method.
The free-body diagram of the four-speed transmission shown in Fig. 2.28 in its first gear is
given in Fig. 2.44. In this free-body diagram, the transmission is divided into six free bodies:
those are first and second sun gears, first and second carriers, and first and second ring gears.
The torques exerted on those free bodies is labeled on them. Here, F1 is the tangential force
between the first sun gear, first planetary gear, and first ring gear. F2 is the tangential force
between the second sun gear, second planetary gear, and second ring gear. Tin is the transmis-
sion input torque, which is applied to the first sun gear. TC1R is the reaction torque exerted on
the first ring gear by the grounding clutch C1R. Tc1r2 is the connection torque of the first carrier
and second ring gear. Tr1c2 is the connection torque of the first ring gear and second carrier.
Tout is the transmission output torque exerted on the second ring gear.
Then the torque equations of those free bodies can be easily obtained:
First ring gear: TC 1R − Tr 1c 2 − F1R1 = 0 (2.54)
First carrier: F1 (S1 + R1) − Tc 1r 2 = 0 (2.55)
Fig. 2.44 Free-body diagram

of 1st gear of the transmission
shown in Section 2.2.2.
35
6293_Book.indb 35 1/11/13 3:52 PM

Chapter 2
First sun gear: Tin − F1S1 = 0 (2.56)
Second ring gear: Tc 1r 2 − Tout + F2 R 2 = 0 (2.57)
Second carrier: Tr 1c 2 − F2 (S 2 + R 2) = 0 (2.58)
Second sun gear: F2S 2 = 0 (2.59)
The equations can be put into the matrix equation format, as shown in Eq. 2.60.
NT = DTin (2.60)
Here,
 S1 0 0 0 0 0 
 
 R1 0 −1 0 1 0 
 −(S1 + R1) 0 0 1 0 0 
N = 
 0 S2 0 0 0 0 
 0 R2 0 −1 0 1 
 
 0 −(S 2 + R 2) 1 0 0 0 
 F1 
 
 F2 
 
Tr 1c 2
T =

 Tc 1r 2 
 
 TC 1R 
 Tout 
 
 1 
 
 0 
 0 
D = 
 0 
 0 
 0 
The solution of Eq. 2.60 is shown in Eq. 2.61. The elements of vector k are the torque ratios of
each node vs. input torque.

T = N −1DTin = kT (2.61)
in
36
6293_Book.indb 36 1/11/13 3:52 PM

Here,
 kF 1/in 
 
 kF 1/in 
 
kTr 1c 2/in
k = 


kTc 1r 2/in
 
 kC 1R /in 
 
 kTout/in 
We can rewrite the result as shown in Eq. 2.62:
 F1   kF 1/in 
   
 F2   kF 1/in 
   
 Tr 1c 2 = kTr 1c 2/in T (2.62)
 Tc 1r 2   kTc 1r 2/in  in
   
 TC 1R   kC 1R /in 
 Tout   
   kTout/in 
It is interesting to note that the first five columns of matrix N are the transposition of the
first five rows of matrix M of Eq. 2.52, and they come from the speed constraint equations of
two planetary gear sets, the two permanent connections, and the grounding clutch. The last
column of matrix N defines the output node. The vector D defines the input node.
From the previous observations, it can be seen that to calculate torque ratios using the matrix
method, one does not need to start from free-body diagrams. All we need to do is to put the
speed constraint equations in the columns of matrix N and add the last column to define the
output node.
As the final words regarding the matrix method that the authors would like to point out,
because the M matrix includes all permanent and clutch connection equations along with speed
constraint equations of planetary gear sets, and because changing the connection equations will
change the topology of planetary gear trains, the matrix method is a convenient tool to synthe-
size planetary gear trains to find a topology that provides the desired ratio and ratio progression.
2.2.6 Gear-shifting mechanics

As discussed in the previous sections, in a planetary gear automatic transmission, a gear ratio
is achieved by engaging the corresponding clutches. To change gear ratio from one to another,
some clutches need to be released from the engaged state, and some clutches need to be
engaged from the released state. If only one clutch needs to be released, and one clutch needs
to be engaged, the gear shifting is called a single-transition shift. If two clutches need to be
released, and two clutches need to be engaged, the shifting is called a double-transition shift.
Since double-transition shifts are difficult to control, most planetary gear automatic transmis-
sions are designed to have only single-transition gear shifts for one- and two-step gear shifts.
37
6293_Book.indb 37 1/11/13 3:52 PM

Chapter 2
To make a smooth gear shift, releasing and applying of clutches has to be synchronized based
on gear-shifting mechanics. This section will discuss the mechanics of gear shifting. Unlike
manual transmissions, automatic transmissions can achieve gear shifts without interrupting
engine power. Therefore, the gear shifts are categorized into four types: power-on up-shift,
power-on down-shift, power-off up-shift, and power-off down-shift. In addition to these four
basic types of gear shifts, there are some other types of variant shifts. For example, the canceled
shift is a shift to return to the original gear during a gear shift; the change mind shift is a shift
in which the target gear is changed due to the driver stepping on or off the accelerator; the
multiple-shift is a shift that consists of two single-transition shifts with smooth transition
from the first one to the second one. The mechanics of all these shifts can be analyzed using
the same methodology. In the following paragraphs, the mechanics of power-on up-shift and
down-shift will be analyzed. Readers can use the same methodology to analyze all other types
of gear shifts.
2.2.6.1 Power-on up-shift

Let’s consider the power-on up-shift first, and look at the shift from first gear to second gear
of the four-speed transmission shown in Fig. 2.28. To accomplish the shift, clutch C1R needs
to be released, clutch C24 needs to be applied, and clutch C123 is kept engaged throughout
the shift. The lever diagram for analyzing the first to second gear shifting is shown in Fig. 2.45.
In this figure, the first gear line and second gear line crossed at the same output speed point.
This is because during a gear shift the output speed can be considered unchanged due to the
large vehicle mass. The input speed decreases from first gear speed level to second gear level
when shifting from first gear to second gear. Also, it should be noted that the slip speed of
C24 is negative when in first gear, because the C24 node on the first gear line is above the zero
speed line. The second lever diagram in Fig. 2.45 shows all torques exerted on the lever. Here,
Tin is transmission input torque, Tout is transmission output torque, TC1R is the torque of the
oncoming clutch C1R, and TC24 is the torque of the off-going clutch C24. The torque equations
2.63–2.66 for discussions of shift mechanics are obtained from this lever diagram using the
principle of moment balancing discussed in Section 2.2.4.
Before the discussions of gear-shifting mechanics, let’s review two concepts mentioned previ-
ously. The first one is the “torque capacity” of a clutch. The torque capacity of a clutch is the
Fig. 2.45 Lever diagram for

analyzing the first to
second gear shift.
38
6293_Book.indb 38 1/11/13 3:52 PM

maximum torque the clutch can transfer; it is controlled by the normal force applied on the
clutch plates. The second concept is the actual “torque” transferred by a clutch. The actual
torque transferred by a clutch equals the torque capacity when the slip speed of the clutch
is non-zero. When the slip speed of the clutch is zero, then the actual clutch torque is the
reaction torque to other torques exerted on the lever.
The power-on up-shift is divided into two phases. The first phase is the torque phase. The
second phase is the inertial phase (the root of this naming convention will become self evident
in a moment). During the torque phase, the clutch torque is transferred from C1R (called the
off-going clutch in 1-2 shift) to C24 (called the oncoming clutch in 1-2 shift). The torque phase
is started from an increase in the torque capacity of clutch C24 from zero (refer to Fig. 2.46).
Because the slip speed of clutch C24 is non-zero, the torque of the clutch equals the torque
capacity of the clutch. The direction of C24 torque Tc24 is positive, because the slip speed of
clutch C24 is negative. As the torque of clutch C24 increases, the torque carried by clutch C1R
will decrease. This can be explained by Eq. 2.63. Equation 2.63 is obtained by applying the
moment-balancing principle about the output node (see the second lever diagram of Fig. 2.46).
Note that the torque carried by clutch C1R is the reaction torque to the input torque and the
torque of clutch C24, and is not affected by the apply force of clutch C1R as long as the torque
capacity created by this clutch-apply force is higher than the clutch torque determined by Eq.
2.63. The output torque at this point can be obtained from Eq. 2.64. Equation 2.64 indicates
that as the torque of clutch C24 increases from zero, the output torque decreases from its first
gear level. As the torque of clutch C24 reaches the level given by Eq. 2.65, the torque on clutch
C1R becomes zero. This means that at this point, the apply force of clutch C1R can be reduced
to zero. The C24 torque level given by Eq. 2.65 is called the critical capacity. The output torque
at this point can be obtained by substituting Eq. 2.65 into Eq. 2.64. The result is shown in Eq.
2.66. This is the second gear output torque level. At this point, the torque transfer from clutch
C1R to clutch C24 is complete, and the torque phase is finished. The traces of output torque,
C1R torque, and C24 torque during the torque phase are shown in Fig. 2.46. The end of the
torque phase is marked as point B on the output torque trace. Note: this discussion assumes
that during the torque phase, the input Tin is maintained constant.
R1  R2 
TC 1R = Tin − TC 24  + 1 (2.63)
S1  S2 
Fig. 2.46 Power-on shift

from first gear to second gear.
39
6293_Book.indb 39 1/11/13 3:52 PM

Chapter 2
R1 + S1 R2
Tout = Tin − TC 24 (2.64)
S1 S2
R1S 2
TC 24 = Tin (2.65)
S1(R 2 + S 2)
R1S 2 + R 2S1 + S1S 2

Tout = Tin (2.66)
S1S 2 + S1R 2
The second phase is the inertial phase. In the inertial phase, the input speed change occurs.
The slip speed of clutch C24 will change from a negative value to zero. The slip speed of clutch
C1R will change from zero to a positive value. The output speed maintains nearly unchanged
due to the large vehicle mass. The input speed decreases from the first gear level to the second
gear level. To achieve this speed change, an extra apply force needs to be added to the clutch
C24 to overcome the inertial torque required for the input speed change. Let’s lump all speed-
changing inertias to the input node, and name it as Iin. Then the torque of clutch C24 required
to achieve the speed change can be obtained from the lever diagram in Fig. 2.47. The result is
shown in Eq. 2.67. Note that in the lever diagram of Fig. 2.47, an inertial torque Iina is added
to the input node, and C1R torque is zero. The presence of the inertial torque is how this
portion of the shift received its name—Inertial Phase. Here, a is the acceleration of the input
speed. Because the input speed decreases from the first gear level to the second gear level, a
has a negative value. As the torque of clutch C24 reaches this new level, the output torque also
increases to a new level, shown in Eq. 2.68. This brings the output torque trace to the point
C, as shown in Fig. 2.46. From this point, the input speed will decelerate with the rate of a.
Eventually, the slip speed of clutch C24 will reach zero, and consequently the input speed will
reach the second gear level. This marks the completion of the input speed change. As the speed
change terminates, the inertia torque will disappear, and the output torque and the torque of
clutch C24 will return to the level shown in Eqs. 2.65 and 2.66. This is the end of the inertial
phase, and marked as point D on the output torque trace in Fig. 2.46. Then the 1-2 power-on
up-shift is complete.
R1S 2
TC 24 = (Tin − I inα ) (2.67)
S1(R 2 + S 2)
R1S 2 + R 2S1 + S1S 2

Tout = (Tin − I inα ) (2.68)
S1S 2 + S1R 2
During the inertial phase of a power-on up-shift, the output torque will increase and then
decrease, as shown by the section C to D in Fig. 2.46. This increasing and decreasing of output
torque creates a shift disturbance. Can this output torque disturbance be eliminated? The
answer is yes. Let’s rewrite Eq. 2.67 to obtain Eq. 2.69. It can be seen from Eq. 2.69 that the
desired input speed deceleration a can be achieved by not only increasing C24 torque, but also
by decreasing input torque Tin. This inertial phase control strategy is shown in Fig. 2.48. In Fig.
2.48, the torque of clutch C24 is maintained at the level defined by Eq. 2.65 during the inertial
phase. The input speed deceleration is achieved by the reduction of input torque. The amount
of input torque reduction is Iina. Using this inertial phase control strategy, the output torque
40
6293_Book.indb 40 1/11/13 3:52 PM

Fig. 2.47 Lever diagram

of inertial phase.
disturbance is eliminated. In practice, the reduction in Tin is achieved by controlling the spark
timing of the engine.
The power-on up-shift control methodology described in Fig. 2.48 provides less output torque
disturbance than the one described in Fig. 2.46. However, the modern integrated powertrain
control technologies have achieved the supremely smooth power-on up-shift, in which the
output torque is maintained unchanged before, during, and after the shift. The details of such
a control strategy will be discussed in a later section where the integrated powertrain controls
are discussed.
S1(R 2 + S 2)
I inα = Tin − Tc 24 (2.69)
R1S 2
To achieve a smooth power-on up-shift, synchronization of applying the oncoming clutch and
releasing the off-going clutch is important. The key point of this synchronization is that when
the torque capacity of the oncoming clutch C24 reaches its critical point defined by Eq. 2.65
(this is also the B point on the output torque trace), the torque capacity of the off-going clutch
C1R has to be reduced to zero. If this happens early, the engine speed will flare, and the output
torque will dip down further. If this happens late, the transmission will be tied up, and the
output torque also will dip down further.
The gear-shifting process and its mechanics can be further understood after the discussion of
dynamic modeling and analysis of planetary gear transmission in the later section.
Fig. 2.48 Power-on shift

from first gear to second gear.
41
6293_Book.indb 41 1/11/13 3:52 PM

Chapter 2
2.2.6.2 Power-on down-shift

A power-on down-shift will be required when a driver provides further step-in to the accelera-
tor pedal to demand more power. A power-on down-shift is the reverse process of a power-on
up-shift. Figure 2.49 illustrates the 2-1 power-on down-shift. The shift starts from the reducing
of the capacity of clutch C24 to a level to cause the slip speed of clutch C24 to increase at a
desired rate. As a result, the transmission input speed also increases at the desired rate, and the
slip speed of clutch C1R approaches zero. This speed-change process is called the inertial phase
of power-on down-shifts. Once the slip speed of C1R reaches zero, the capacity of clutch C1R
is rapidly increased to its full level, and the capacity of clutch C24 is rapidly decreased to zero.
This capacity transfer process is the torque phase of power-on down-shifts. The end of the
torque phase marks the completion of a power-on down-shift.
In the previous discussions of gear-shift mechanics of power-on up-shifts and down-shifts, the
lever diagram is used as a tool to understand the direction of speed and torque of each node,
and to derive torque equations. Although we focused on single-transition power-on up and
down shifts, methodology can be used to analyze the mechanics of other types of gear shifts
such as power-off shifts, double-transition shifts, canceled shifts, etc. The emphasis is on the
methodology. The readers are encouraged to review and comprehend the methodology used in
the discussions.
Homework 1 for Section 2.2

1. Perform static analysis of the automatic transmission in Fig. 2.50 using the
algebraic method, lever method, and matrix method. Calculate input-output speed
ratio of each gear; calculate input torque ratio of engaged clutches of each gear.
Here, R1/S1=2.38, R2/ S2=1.76.
Fig. 2.49 Power-on down-shift from second gear to first gear.
42
6293_Book.indb 42 1/11/13 3:52 PM


of a four-speed planetary
automatic transmission.
Clutching Chart
Gear C13R C12 C234 C4 CR
1st gear X X
2nd gear X X
3rd gear X X
4th gear X X
Reverse gear X X
Speed and Torque Ratio Table

Torque Ratio
Speed
Gear Ratio C13R C12 C234 C4 CR
1st gear win/wout TC13R /Tin TC12/Tin
2nd gear win/wout TC12/Tin TC234/Tin
3rd gear win/wout TC13R /Tin TC234/Tin
4th gear win/wout TC234/Tin TC4/Tin
Reverse gear win/wout TC13R /Tin TCR/Tin
2. Use the analysis method of your choice to calculate the input-output speed ratio of each
gear, and calculate the input torque ratio of the engaged clutches of each gear of the trans-
mission system shown in Figs. 2.51 and 2.52.
Number of Teeth
# S SP, P1 RP, P2 R
1 37 27 25 97
2 47 25
3 49 20 89
43
6293_Book.indb 43 1/11/13 3:52 PM

Chapter 2
Fig. 2.51 Stick diagram of a six-speed planetary automatic transmission.
Fig. 2.52 Lever diagram of a six-speed planetary automatic transmission.
Gear C1234 F1 CB26 C35R C456 CBR
Rev X X
1st X X
2nd X X
3rd X X
4th X X
5th X X
6th X X
3. Develop the torque equations and speed equations of the Ravigneaux gear set shown in
Fig. 2.53.
44
6293_Book.indb 44 1/11/13 3:52 PM


a Ravigneaux gear set.
of
2.2.7 Frictional clutches and their mathematical models

Prior to the discussions of dynamic analysis of planetary gear automatic transmissions, this
section provides some insight into the mathematical modeling of frictional clutches that
is needed in dynamic modeling of automatic transmissions. As described in the previous
sections, to achieve a desired gear ratio, the corresponding elements of a planetary gear train
need to be connected to input, output, or ground. Frictional clutches are used to accomplish
this in automatic transmissions. Figure 2.54 is a cross-section view of a frictional clutch. A
hydraulic piston presses on two sets of friction plates. One set of friction plates is splined on
shaft A (rotationally keyed to the shaft but able to slide axially), and the other set is splined
on shaft B. The spline allows the frictional plates to move axially while transmitting torque.
The hydraulic cylinder and piston rotate with shaft A. The product of hydraulic pressure and
piston area determines the normal force pressing on two sets of friction plates, and therefore
determines the transmittable torque (torque capacity) of the clutch. The diagram symbol for
frictional clutches is shown in Fig. 2.55.
In dynamic models of automatic transmissions (as we will discuss in the following sections),
the frictional clutches can be mathematically modeled using a hyperbolic tangent function, as
shown in Eq. 2.70.
Tc = µ ⋅ n ⋅ r (A ⋅ p − Fs )tanh(Δω / α ) = Tcap tanh(Δω / α ) (2.70)
Fig. 2.55 Symbol for

Fig. 2.54 Cross section of frictional clutches. the frictional clutches.
45
6293_Book.indb 45 1/11/13 3:52 PM

Chapter 2
Here, Tc is the clutch torque, m is the friction coefficient, n is the number of the friction
surface, A is the area of the clutch piston, r is the radius of the clutch plate, p is the hydraulic
pressure, tanh(·) is the hyperbolic tangent function, a is the scaling factor of the hyperbolic
tangent function, Δw is the differential speed of the two shafts, and Fs is the clutch return
spring force. Tcap = (m · n · r · A · p – Fs) is called the clutch capacity. Figure 2.56 shows the
graphics of the clutch model of Eq. 2.70. Note that the value of a determines the slope of the
function near zero slip speed.
The hyperbolic tangent function clutch model of Eq. 2.70 is simple and easy to use, but it has a
couple of minor problems. First, to carry torque, the clutch slip speed has to be non-zero. This
means that when the clutch is in the engaged state, there will be a small slip speed across the
clutch in this mathematical model. Second, although this slip speed can be reduced by using
a small scaling factor a, a small scaling factor could make the simulation model run slow or
could make the model unstable.
An alternate model of frictional clutches is shown in Fig. 2.57. The model uses PI (Propor-
tional and Integral) control techniques to calculate clutch torque. Here, Ki is the integral gain,
and Kp is the proportional gain; they are used to calibrate the model. The model is called the
kpki-model.
One can build Simulink models as shown in Figs. 2.58 and 2.59 to compare these two fric-
tional clutch models. The physical system model of those Simulink models is shown in Fig.
2.60. In Fig. 2.58, the submodel labeled “kpki model” is the model shown in Fig. 2.57. In Fig.
2.59, the submodel labeled “tanh model” is a MATLAB function of Eq. 2.70.
Fig. 2.56 Graphics of a hyperbolic

tangent function model of clutches.
Fig. 2.57 Alternate model of frictional clutches: kpki-model.
46
6293_Book.indb 46 1/11/13 3:52 PM

Fig. 2.58 Simulink model for testing the kpki clutch model.
Fig. 2.59 Simulink model for testing the tanh clutch model.
Fig. 2.60 Clutch model test system.
2.2.8 Dynamic equations of simple planetary gear sets

The next four sections are devoted to dynamic analysis of planetary gear trains. First, let’s
consider the simple planetary gear set, and then expand the discussion to planetary gear trains.
2.2.8.1 Lagrange equation of the system

Because there are three speed elements, ws, wr, and wc, and one speed constraint equation
shown in Eq. 2.71, a simple planetary gear set has two degrees of freedom. Let’s take those three
speeds as generic coordinates of the system. The dynamic behavior of a simple planetary gear
47
6293_Book.indb 47 1/11/13 3:52 PM

Chapter 2

of a simple planetary gear set.
set can be described by three differential equations and one speed constraint equation. The
three differential equations can be obtained from the free-body diagrams shown in Fig. 2.61.
ωs S + ωr R = ωc (R + S ) (2.71a)
ω s S + ωr R = ωc (R + S ) (2.71b)
In Fig. 2.61, Ir, Is, and Ic are the inertia of the ring gear, sun gear, and carrier, respectively. F is
the contact force between the gear teeth. Tr, Ts, and Tc are the torques exerted on the ring gear,
sun gear, and carrier, respectively. There are total three free bodies in a planetary gear set. They
are the ring gear, sun gear, and carrier. From the free-body diagram, the dynamic equation of
each free body can be obtained, as shown in Eqs. 2.72–2.74.
ω r I r = Tr − F ⋅ R (2.72)
ω s I s = Ts − F ⋅ S (2.73)
ω c I c = F ⋅ R + F ⋅ S − Tc (2.74)
Equations 2.71–2.74 can be rearranged into a matrix form, as shown in Eq. 2.75. This is the
Lagrange equation of a simple planetary gear set. By left-multiplying the equation with the
inverse of M, the vector W with accelerations of sun, ring, and carrier and the internal force F
as its components can be obtained.
The Lagrange equation of a simple planetary gear set has an interesting structure. The upper-
left of matrix M is a three by three diagonal matrix with all inertias of the system on its main
diagonal. The lower-left of matrix M is a one by three matrix that corresponds to the speed
constraint equation of a planetary gear set. The upper-right of matrix M is a three by one
matrix, which is the transposition of the lower-left matrix. The lower-right of matrix M is a
one by one zero matrix. The matrix B after the equal sign defines which external torque acts on
which inertia. Vector T has all external torque values as its components. The internal force F is
called the Lagrange multiplier in some literatures.
MΩ = BT (2.75)
48
6293_Book.indb 48 1/11/13 3:52 PM

Here,
 Ir 0 0 R   ω r 
   
 0 Is 0 S   ω s 
M = , Ω = 
 0 0 Ic −(R + S )   ω c 
 R S −(R + S ) 0   F 
 
 1 0 0   T 
   r 
0 1 0 
B = , T =  Ts 
 0 0 −1   

 0 0 0   Tc 
2.2.8.2 Reduced-order equation of the system

The Lagrange equation is an efficient way to describe the dynamic model of a simple planetary
gear set and also planetary gear trains. However, as will be demonstrated later, the following
mathematical procedure is also useful. In this procedure, by eliminating the internal variable F
and applying the speed constraint equation (Eq. 2.71), the previous three differential equations
can be reduced to two differential equations, as shown in Eqs. 2.76 and 2.77.
 S 2Ir Ic  (R + S )SI r RSI c

ω s  I s +  = Ts − Tc − Tr (2.76)
 (R + S )2
I r + R 2
I c  (R + S )2
I r + R 2
I c (R + S )2
Ir + R 2Ic
 (R + S )2 I r I s  (R + S )RI s (R + S )SI r
ω c  I c + 2  = Tr 2 + Ts 2 − Tc (2.77)
 S Ir + R I s 
2
S Ir + R I s
2
S Ir + R 2I s
The dynamic behavior of the sun gear and carrier are described by these two differential
equations. The ring gear speed and acceleration can be obtained from Eq. 2.71. The reduced-
order system model has two independent differential equations; this is because the system
has three speed coordinates and one speed constraint equation, and therefore has two degrees
of freedom.
2.2.9 Dynamic equations and simulation model of

planetary gear trains
This section presents the dynamic equations and Simulink model of planetary gear automatic
transmissions. Let’s use the four-speed planetary automatic transmission shown in Fig. 2.62 as
an example for our discussions. The lever diagram of the transmission is shown in Fig. 2.63.
2.2.9.1 Lagrange equation of the system

The dynamic equations of the previous transmission system can be developed from its free-
body diagram shown in Fig. 2.64. There are five free bodies. The first free body is the first
49
6293_Book.indb 49 1/11/13 3:52 PM

Chapter 2
Fig. 2.62 Stick diagram of a four-speed planetary gear automatic transmission.
Fig. 2.63 Lever diagram of a four-speed planetary gear automatic transmission.
Fig. 2.64 Free-body diagram of a four-speed planetary gear automatic transmission.
sun gear, SG1, with inertial of IS1 and angular velocity of wS1. The second free body is the first
carrier and second ring gear, C1+RG2, with inertial of IC1R2O and angular velocity of wC1R2.
The third free body is the first ring gear and second carrier, RG1+C2, with inertial of IC2R1 and
angular velocity of wC2R1. The fourth free body is the second sun gear, SG2, with inertial of IS2
and angular velocity of wS2. The fifth free body is the turbine shaft with inertial of It and angular
velocity of wt. The dynamic equations for these five free bodies are shown in Eqs. 2.78–2.82.
50
6293_Book.indb 50 1/11/13 3:52 PM

TC123 − F1S1 = ω S1 I S1
(2.78)
F1S1 + F1 R1 − F2 R2 − TO = ω C 1R 2 I C 1R 2O
(2.79)
TC1R + TC 34 − F1 R1 + F2 R2 + F2 S2 = ω R1C 2 I R1C 2 (2.80)
TC 24 + TCR − F2 S2 = ω S 2 I S 2 (2.81)
Tin − TC123 − TCR − TC 34 = ω t I t (2.82)
Here, TCXXX is the torque of the clutch CXXX. F1 and F2 are the contact forces between the
sun gear, pinion gear, and ring gear of two planetary gear sets, respectively. The speed con-
straint equations for the first and second planetary gear sets are shown in Eqs. 2.83 and 2.84,
respectively.
ωR1C 2 R1 + ωS1S1 = ωC 1R 2 (R1 + S1 ) (2.83)
ωC 1R 2 R2 + ωS 2S2 = ωR1C 2 (R2 + S2 ) (2.84)
Arranging Eqs. 2.78–2.84 into a matrix form results in the Lagrange model of the system, as
shown in Eq. 2.85. Like the Lagrange equation of a simple planetary, matrix M in Eq. 2.85 has
some interesting features. The upper-left of matrix M is a five by five diagonal matrix with all
inertias of the system on its main diagonal. The lower-left of matrix M is a two by five matrix
that corresponds to the two speed constraint equations. The upper-right is a five by two matrix
that is the transposition of the lower-left matrix. The lower-right is a two by two zero matrix.
The column vector Ω is the state variable vector, which consists of the acceleration variables
of all free bodies and internal forces F1 and F2. The vector T consists of all clutch torques and
external torques. The clutch torques can be obtained using the clutch model given in Section
2.2.7 with corresponding shaft speeds. The matrix B describes which torque is exerted on
which inertia in what direction.
The Simulink model of the four-speed transmission based on the Lagrange equation, Eq. 2.85,
is shown in Fig. 2.65. In the Simulink model, the vector bar T represents the torque vector T.
The vector bar Ω represents the state vector Ω. The MATLAB Function F in the Simulink model
solves Eq. 2.85 by left multiplying the inverse of matrix M. The functions T_c123, T_c1r, T_
c34, T_c24, and T_cr at the right side of the model calculate clutch torques. The inputs to these
functions are the two speeds of the clutch shafts, and clutch capacities labeled by T_cap_c123,
T_cap_c1r, T_cap_c34, T_cap_c24, and T_cap_cr. The model can be used to simulate any
gear shift of the transmission, including any single-transition up and down shifts as well as
any double-transition shifts. Analysis of gear shifts using dynamic simulation models will be
discussed in a later section as we complete the introduction of dynamic modeling of planetary
gear transmissions.
MΩ = BT (2.85)
51
6293_Book.indb 51 1/11/13 3:52 PM

Chapter 2
Fig. 2.65 Simulink model of a four-speed automatic transmission based on its Lagrange equation.
Here,
 I s1 0 0 0 0 S1 0 
 
 0 I c 1r 2o 0 0 0 −(R1 + S1 ) R2 
 
 0 0 I r 1c 2 0 0 R1 −(R2 + S2 ) 
M = 0 0 0 Is2 0 0 S2 
 
 0 0 0 0 It 0 0 
 
 S1 −(R1 + S1 ) R1 0 0 0 0 
 0 R2 −(R2 + S2 ) S2 0 0 0 
 
  TC 123 
ω s1     
  1 0 0 0 0 0 0
 ω c 1r 2     To 
 0 −1 0 0 0 0 0   
  TC 1R 
 ω r 1c 2   0 0 1 1 0 0 0  
Ω = ω s 2  B = 0 0 0 0 1 1 0  T = TC 34 
     
 ω t   −1 0 0 −1 0 −1 1   TC 24 
   0 0 0 0 0 0 0   
F1     TCR 
  0 0 0 0 0 0 0 
 F 2   Tin 

52
6293_Book.indb 52 1/11/13 3:52 PM

2.2.9.2 Reduced-order equation of the system

Applying the same mathematical procedure as we did for the simple planetary gear set case,
the dynamic equations, Eqs. 2.78–2.81, can be reduced to two differential equations, as shown
in Eqs. 2.86 and 2.87. The Simulink model based on these equations is shown in Fig. 2.66.
ω C 1R 2 IC 1R 2 = TC 123 K C 123−C 1R 2 + (TCR + TC 24 )K C 24 −C 1R 2 + (TC 1R + TC 34 )K C 34 −C 1R 2 − TO (2.86)
ω R1C 2 IR1C 2 = (TC 1R + TC 34 ) + (TCR + TC 24 )K C 24−R1C 2 − TC 123 K C 123−R1C 2 − TO K O −R1C 2 (2.87)
Here,
IC 1R 2 = I C 1R 2O +
I R1C 2 I S1 ( R1 +S1 2
S1 ) + I R1C 2 I S 2 ( ) +I R2 2
S2 I
S1 S 2 ( S1S2 )
R1S2 +S1R2 +S1S2 2
I R1C 2 + I S1 ( ) +I (
R1 2
S1 S2 S2 )
R 2 +S 2 2
IR1C 2 = I R1C 2 +
I C 1R 2O I S1 ( ) R1 2
S1 + I C 1R 2O I S 2 ( S2 )
R 2 +S 2 2
+ I S1I S 2 ( )
R1S2 +S1R2 +S1S2 2
S1S2
I C 1R 2O + I S1 ( R1 +S1 2
S1 ) + IS2 ( )R2 2
S2
K C 123−C 1R 2 =
I R1C 2 R1 +S1
S1 + IS2 ( R 2 +S 2
S2 )( R1S2 +S1R2 +S1S2
S1S2 )
I R1C 2 + I S1 ( )
R1 2
S1 + IS2 ( R 2 +S 2 2
S2 )
K C 24 −C 1R 2 =
−I R1C 2 R2
S2 + I S1 ( )( R1
S1
R1S2 +S1R2 +S1S2
S1S2 )
I R1C 2 + I S1 ( ) +I (
R1 2
S1 S2
R 2 +S 2 2
S2 )
I S1 R1 (RS12+S1 ) + I S 2 R 2 ( R 2 +S 2 )
S22
K C 34 −C 1R 2 = 1
I R1C 2 + I S1 ( ) R1 2
S1 + IS2 ( R 2 +S 2 2
S2 )
K C 24 −R1C 2 =
I C 1R 2O R 2 +S 2
S2 + I S1 ( R1 +S1
S1 )( R1S2 +S1R2 +S1S2
S1S2 )
I C 1R 2O + I S1 ( R1 +S1 2
S1 ) + IS2 ( ) R2 2
S2
K C 123−R1C 2 =
I C 1R 2O R1
S1 + IS2 ( )( R2
S2 ) R1S2 +S1R2 +S1S2
S1S2
I C 1R 2O + I S1 ( ) +I ( )
R1 +S1 2
S1 S2
R2 2
S2
I S1 R1 (RS12+S1 ) + I S 2 R 2 ( R 2 +S 2 )
S22
K O −R1C 2 = 1
I R1C 2O + I S1 ( S1 )
R1 +S1 2
+ IS2 ( ) R2 2
S2
53
6293_Book.indb 53 1/11/13 3:52 PM

Chapter 2
Fig. 2.66 Simulink model of a four-speed automatic transmission based on the reduced-order equations.
2.2.9.3 Directly obtain Lagrange equation of a system

Equations 2.75 and 2.85 are Lagrange equations of a simple planetary gear set and a four-speed
automatic transmission, respectively. As mentioned earlier, they all have an intersecting struc-
ture, as summarized by Eq. 2.88. Here, J is the diagonal matrix with all system inertias on its
main diagonal, Φ is the constraint equation matrix, ∆ is the acceleration vector of all inertias,
F is the internal force and torque vector, Π is the external torque apply matrix, and T is the
external torque vector.
    
 J Φ  Δ  =  Π  T
T
(2.88)
 Φ 0  F   0 
With the previous understandings of the structure of the Lagrange equation of a planetary gear
automatic transmission system, we can determine the Lagrange equation of the system shown
in Fig. 2.62 directly, without going through free-body diagrams. The result is given in Eq. 2.89.
Note: in Eq. 2.89 there are a total of eight inertia elements. They are the inertias of sun gear,
carrier, and ring gear of two planetary gear sets plus input shaft inertia and output shaft
inertia. Therefore, three additional constraint equations represented by the first three rows of
the constraint matrix are necessary. The first row [0 1 0 0 0 –1 0 0] represents the connection
of the first carrier C1 and second ring gear R2. The second row [0 0 1 0 –1 0 0 0] represents
the connection of the first ring gear R1 and second carrier C2. The third row [0 0 0 0 0 1 0
–1] represents the connection of the second ring gear R2 and the output shaft. The Lagrange
multipliers F1, F2, and F3 are the internal torque of these three connections, respectively. The
54
6293_Book.indb 54 1/11/13 3:52 PM

Lagrange multipliers F4 and F5 are the contact force of gear teeth of first and second planetary
gear set, respectively. The matrix B describes how the external torque (including the torque of
clutches) is applied to these inertia elements.
Although both Eq. 2.89 and Eq. 2.85 are the Lagrange equations of the same transmission
system shown in Fig. 2.62, they have different sizes. This is because in Eq. 2.85 we treated the
permanently connected elements as single inertia; but contrastingly in Eq. 2.89 we consider
them as separate inertias, and use the constraint equation to describe the connections.
MΩ = BT (2.89)
Here,
 I s1 0 0 0 0 0 0 0 0 0 0 S1  0
 
 0 Ic1 0 0 0 0 0 0 1 0 0 −(R1 + S1) 0 
 
 0 0 Ir1 0 0 0 0 0 0 1 0 R1 0 
 0 0 0 Is2 0 0 0 0 0 0 0 0 S2 
 
 0 0 0 0 Ic 2 0 0 0 0 −1 0 0 −(R 2 + S 2) 
 
 0 0 0 0 0 Ir 2 0 0 −1 0 1 0 R2 
M = 0 0 0 0 0 0 It 0 0 0 0 0 0 
 
 0 0 0 0 0 0 0 I o 0 0 −1 0 0 
 0 1 0 0 0 −1 0 0 0 0 0 0 0 
 
 0 0 1 0 −1 0 0 0 0 0 0 0 0 
 0 0 0 0 0 1 0 −1 0 0 0 0 0 
 
 S1 −(R1 + S1) R1 0 0 0 0 0 0 0 0 0 0 
 0 0 0 S 2 −(R 2 + S 2) R 2 0 0 0 0 0 0 0 
 ω s1 
 
 ω c 1   1 0 0 0 0 0 0 
   
ω r 1   0 0 0 0 0 0 0 

  
 ω s 2  
0 0 1 1 0 0 0
 TC 123 
  0 0 0 0 1 1 0  
 ω c 2     To 
   0 0 0 0 0 0 0   
 ω r 2   0 0 0 0 0 0 0   TC 1R 
 
Ω = ω t  B = −1 0 0 −1 0 −1 1  T = TC 34 
   
 ω o   0 −1 0 0 0 0 0   TC 24 
    
F1  
0 0 0 0 0 0 0
  TCR 
  0 0 0 0 0 0 0
 F2     Tin 
   0 0 0 0 0 0 0  
 F3   
 0 0 0 0 0 0 0 
 F4 
   0 0 0 0 0 0 0 
 F5 
55
6293_Book.indb 55 1/11/13 3:52 PM

Chapter 2
2.2.9.4 Connect the transmission model to the vehicle and engine

Now let’s consider how to connect the simulation model of the four-speed automatic transmis-
sion we developed to the outside world.
At the input side, the transmission input shaft is connected to the turbine shaft of the torque
converter; therefore, the transmission input torque Tin in the four-speed transmission model
should be the turbine torque Tt, and the inertia of the input element should be turbine shaft
inertia It. The calculation of turbine torque has been discussed in the previous Section 2.1.2.
At the output side, the transmission output shaft is connected to the vehicle wheel through a
final drive gear with a gear ratio of FR. If we only want to model the inertia load of the vehicle,
then the output torque To in the previous model should be zero, and the inertia of the output
element Ic1r2o should be as given in Eq. 2.90.
R 
2
I c 1r 2o = M v  w  (2.90)
 FR 
Here, Mv is the vehicle mass, Rw is the vehicle wheel radius, and FR is the final drive gear ratio.
If we also would like to model the vehicle drag load, then the output torque To should be as
shown in Eq. 2.91, and the inertia of the output element Ic1r2o should be as given in Eq. 2.90.
R  R 
2 2
Rw
To = f 0 M v g + f 1  w  ωc 1r 2 + f 2  w  ωc21r 2 (2.91)
FR  FR   FR 
Here, f0 is the vehicle rolling resistance coefficient, f1 is the vehicle linear resistance coefficient,
f2 is the vehicle aerodynamic drag coefficient, g is the acceleration of gravity, and wc1r2 is the
speed of the transmission output shaft.
To model the driveline compliance, the output torque To should be as given in Eq. 2.92, and the
inertia of the output element Ic1r2o should be equal to the inertia of transmission output shaft.
The compliance of the driveline is modeled with a spring element between the transmission
output shaft and the input shaft of the final drive gear.
To = K a (θo − θa ) (2.92)
In Eq. 2.92, Ka is the spring constant, qo is the angular displacement of the transmission output
shaft, and qa is the angular displacement of the input shaft of the final drive gear; qo can be
obtained by integrating the transmission output shaft speed, and qa can be obtained by inte-
grating the speed of the input shaft of the final drive gear. The speed of the input shaft of the
final drive gear is given by the following equation.
V
ωa = FR ⋅ (2.93)
Rw
.
Here, V is the vehicle speed; it is determined by the following differential equation.
V = (To − f 0 M v g − f 1 ⋅V − f 2 ⋅V 2 ) / M v (2.94)
56
6293_Book.indb 56 1/11/13 3:52 PM

Note that in the previous model it is assumed there is no vehicle wheel slip. If one would like to
capture the wheel slip also, then the tire model must be included.
2.2.10 Generic dual-clutch model

The dynamic models developed in Section 2.2.9 include all clutches and inertias of the four-
speed automatic transmission. If our focus is on studying a single-transition gear shift, it is not
necessary to have all inertias and clutches in the model. Because during the gear shift only two
clutches (oncoming and off-going) are involved, we will look at the dynamic model including
only these clutches.
Let’s consider the 2-3 shift of the transmission presented in Fig. 2.62. The stick diagram and
lever diagram are then reduced to the ones shown in Fig. 2.67 and Fig. 2.68, respectively.
Using the free-body method, the four dynamic equations of the system, Eqs. 2.95–2.98,
are obtained.
Tin − Tc 34 − F1S1 = ω in I in (2.95)
F1S1 + F1 R1 − F2 R2 − TO = ω C 1R 2 I C 1R 2O (2.96)
TC 34 − F1 R1 + F2 R2 + F2 S2 = ω R1C 2 I R1C 2 (2.97)
TC 24 − F2 S2 = ω S 2 I S 2 (2.98)
Because, in comparison with the inertias of the input and output shaft, inertia IS2 and IR1C2 can
be ignored, let’s make IS2 = 0 and IR1C2 = 0. Then the four dynamic equations are reduced to
two equations, Eqs. 2.99 and 2.100.
Fig. 2.67 Stick diagram of 2-3 shift of a four-

speed planetary gear automatic transmission.
Fig. 2.68 Lever diagram of 2-3 shift of a four-

speed planetary gear automatic transmission.
57
6293_Book.indb 57 1/11/13 3:52 PM

Chapter 2
R1 + S1 (R + S )S
Tin − Tc 34 − Tc 24 2 2 1 = ω in I in (2.99)
R1 S2 R1
R1 + S1 (R + S )(R + S ) − R1R2
Tc 34 + Tc 24 1 1 2 2 − TO = ω O I O (2.100)
R1 R1S2
Here, wO is the output speed and equals wC1R2, IO is the inertia of the output shaft, win is the
input speed, and Iin is the inertia of the input shaft. Tin is the input torque, To is the output
torque, Tc34 is the torque of clutch C34, Tc24 is the torque of clutch C24. R1, S1, R2, and S2 are
the number of teeth of the ring gear and sun gear of the two planetary gear sets, respectively.
To calculate Tc34 and Tc24, the slip speeds of clutches C23 and C24 are necessary. Using the
speed constraint equations, Eqs. 2.83 and 2.84, and letting wS1=win and wC1R2=wo, the slip
speed of clutches C23 and C24 can be expressed in terms of input speed win and output speed
wO. The results are shown in Eqs. 2.101 and 2.102.
R1 + S1 R +S
ΔωC 34 = ωin − ωR1C 2 = ωin − ωO 1 1 (2.101)
R1 R1
S1 (R2 + S2 ) (R + S )(R + S ) − R1R2

ΔωC 24 = 0 − ωS 2 = ωin − ωO 1 1 2 2 (2.102)
R1S2 R1S2
Equations 2.99–2.102 can be written in the generic dual-clutch model form, as shown in Eqs.
2.103–2.106.
Tin − Tc 34kTin /Tc 34 − Tc 24kTin /Tc 24 = ω in I in (2.103)
Tc 34kTo /Tc 34 + Tc 24kTo /Tc 24 − TO = ω O I O (2.104)
ΔωC 34 = ωin − ωR1C 2 = ωinkTin /Tc 34 − ωO kTo /Tc 34 (2.105)
ΔωC 24 = 0 − ωS 2 = ωinkTin /Tc 24 − ωO kTo /Tc 24 (2.106)
Here,
Tc 34 = Tcap _ c 34 tanh(Δωc 34 / ac 34 ) (2.107)
Tc 24 = Tcap _ c 24 tanh(Δωc 24 / ac 24 ) (2.108)
R1 + S1 (R + S )S
Here, kTin /Tc 34 = is the input torque ratio of clutch C34, kTin /Tc 24 = 2 2 1 is the
R1 S2 R1
R +S
input torque ratio of clutch C24, kTo /Tc 34 = 1 1 is the output torque ratio of clutch C34, and
R1
(R + S )(R + S ) − R1R2
kTo /Tc 24 = 1 1 2 2 is the output torque ratio of clutch C24. Note: the second gear
R1S2
ratio r2 = kTo /Tc 24 / kTin /Tc 24 ; and third gear ratio r3 = kTo /Tc 34 / kTin /Tc 34 .
58
6293_Book.indb 58 1/11/13 3:52 PM

The Simulink model of such a dual-clutch system is shown in Fig. 2.69. This generic dual-
clutch model also can be considered as the model of the physical system shown in Fig. 2.70.
The reason the dual-clutch model is referred to as the generic dual-clutch model is because the
model can be used to simulate any single-transition gear shift. To simulate a specific gear shift,
all we need to do is to obtain the input and output torque ratios of two clutches using one of
the static analysis methods given in the previous sections, and enter them into the model.
Figures 2.71–2.73 show the simulation results of a power-on up-shift using the generic dual-
clutch model. In the simulations, the transmission input torque is kept constant. Figure 2.71
is the simulation result in which the off-going clutch torque capacity is dropped to zero when
Fig. 2.69 Simulink model of a generic dual-clutch system.
Fig. 2.70 Generic dual-clutch system.
59
6293_Book.indb 59 1/11/13 3:52 PM

Chapter 2
Fig. 2.71 Simulation result of a perfect power-on up-shift.
the oncoming clutch reaches critical capacity, so it is a perfect power-on up-shift. Figure 2.72 is
the simulation result in which the off-going clutch torque capacity is dropped to zero too early;
it results in more output torque drop and engine speed flare. Figure 2.73 is the simulation of
dropping the off-going clutch capacity to zero too late, and it results in tie-up, and more output
torque drop.
2.2.11 Matrix dual-clutch model

The dual-clutch model of a specific gear shift of a planetary gear transmission can also be
derived using the matrix expression. Let’s consider the 2-3 shift of the transmission shown in
Fig. 2.62 again. Because the clutch C123 is locked during the 2-3 shift, its torque TC123 becomes
an internal constraint torque, and an additional speed constraint equation, as shown in Eq.
2.109, is needed. Also, because clutches C1R and CR are always in the released state, we have
TC1R = TCR = 0.
ωs1 − ωr = 0 (2.109)
With these additional constraints, Eq. 2.85 becomes the one shown in Eq. 2.110. Note that now
the torque vector T only has two clutch torques. The matrix M has an additional row (last row)
and an additional column (last column), due to the new speed-constraint equation, Eq. 2.109.
60
6293_Book.indb 60 1/11/13 3:52 PM

Fig. 2.72 Simulation result of releasing off-going clutch too early.
Fig. 2.73 Simulation result of releasing off-going clutch too late.
61
6293_Book.indb 61 1/11/13 3:52 PM

Chapter 2
The vector Ω has an additional element TC123. Similarly, as shown in the early section, based on
this matrix equation, a Simulink model can be built to simulate the 2-3 shift.
MΩ = BT (2.110)
Here,
 I s1 0 0 0 0 S1 0 1 
 
 0 I c 1r 2o 0 0 0 −(R1 + S1 ) R2 0 
 
 0 0 I r 1c 2 0 0 R1 −(R2 + S2 ) 0 
 0 0 0 Is2 0 0 S2 0 
M = 
 0 0 0 0 It 0 0 −1 
 
 S1 −(R1 + S1 ) R1 0 0 0 0 0 
 0 R2 −(R2 + S2 ) S2 0 0 0 0 
 
 1 0 0 0 −1 0 0 0 
 ω s1 
   0 0 0 0 
 ω c 1r 2   
   1 0 0 0   T0 
 ω r 2c 1   0 1 0 0   
  
ω s 2  0 0 1 0  Tc 34 
Ω =  B =  T = 
 ω t   0 −1 0 1   Tc 24 
    
F1  
0 0 0 0
  Tin 
 0 0 0 0
 F2   
  0 0 0 0 
 TC 123 
2.2.12 Inertia balancing

Another application of dynamic analysis of planetary gear trains is inertia balancing. Let’s
consider the gear train shown in Fig. 2.74. When the second clutch is released, the gear train
Fig. 2.74 Stick and lever

diagram of a 1-2 shift gear train.
62
6293_Book.indb 62 1/11/13 3:52 PM

will be in the first gear (by virtue of the Lo Roller passive torque reaction). When the second
clutch is engaged, the gear train will be in the second gear. A concern here is that, during gear
shifting from first gear to second gear, if there are disturbances in the second clutch pressure,
how they will affect shift feel.
Let’s establish the dynamic equations first. The free-body diagram of the system is shown
in Fig. 2.75.
The four free-body equations are provided in Eqs. 2.111–2.114.
Tc − F1S1 = ω s1 I s1 (2.111)
F1 (R1 + S1 ) − F2 R2 = ω c1r 2 I c1r 2 (2.112)
F2 (R2 + S2 ) − F1 R1 − To = ω c 2r1 I c 2r1o (2.113)
Ti − F2 S2 = ω s 2 I s 2 (2.114)
The two constraint equations are given in Eqs. 2.115 and 2.116.
ωc 2r 1R1 + ωs1S1 = ωc 1r 2 (R1 + S1 ) (2.115)
ωc 1r 2 R2 + ωs 2S2 = ωc 2r 1 (R2 + S2 ) (2.116)
Therefore, the system has two degree of freedom. After eliminating the two internal forces F1
and F2, and applying the two speed constraint equations, differential equations Eq. 2.117 and
Eq. 2.118 are obtained.

ωc 2r 1 I 11 − ωs 2 I 12 = Tc a1 − To (2.117)
−ω c 2r 1 I 21 + ω s 2 I 22 = Ti − Tc a2 (2.118)
Here,
R1S2 + S1R2 + S1S2
a1 = (2.119)
S1R2

a 1-2 shift gear train.
of
63
6293_Book.indb 63 1/11/13 3:52 PM

Chapter 2
(R1 + S1 )S2
a2 = (2.120)
S1R2
R2 + S2 2 R S +S R +S S
I 11 = I c 2r 1o + ( ) I c 1r 2 + ( 1 2 1 2 1 2 )2 I s1 (2.121)
R2 S1R2
(R2 + S2 )S2 (R S + S R + S S )(R + S )S

I 12 = I 21 = 2
I c 1r 2 + 1 2 1 2 1 22 1 1 2 I s1 (2.122)
R2 (S1R2 )
S   (R + S )S 
2 2
I 22 = I s 2 +  2  I c 1r 2 +  1 1 2  I s1 (2.123)
 R2   S1R2 
Equations 2.117 and 2.118 can be decoupled with respect to ω c 2r1 and ω s 2 , as shown in Eqs.
2.124 and 2.125.
−Tc (a2 I 11 − a1 I 21 ) + Ti I 11 − To I 21
ω s 2 = (2.124)
I 11 I 22 − I 12 I 21
Tc (a1 I 22 − a2 I 12 ) + Ti I 12 − To I 22
ω c 2r 1 = (2.125)
I 11 I 22 − I 12 I 21
In Eqs. 2.124 and 2.125,
I 11 I 22 − I 12 I 21 =
 (R + S )S  S 
2 2
I c 2r 1 I s 2 + I c 2r 1 I s1  1 1 2  + I c 2r 1 I c 1r 2  2  + (2.126)
 S1R2   R2 
R +S  S R  R S +S R +S S 
2 2 2
I c 1r 2 I s 2  2 2  + I c 1r 2 I s1  2 1  + I s1 I s 2  1 2 1 2 1 2 
 R2   S1R2   S1R2 
(R1 + S1 )S2 (R + S )S R
a2 I 11 − a1 I 21 = I c 2r 1 + I c 1r 2 2 2 2 2 1 (2.127)
S1R2 S1R2
R1S2 + S1R2 + S1S2 S 2R

a1 I 22 − a2 I 12 = I s 2 + I c 1r 2 2 21 (2.128)
S1R2 S1R2
Equations 2.124 and 2.125 can be rewritten as shown in Eqs. 2.129 and 2.130.
ω s 2 Is 2 = Ti − Tc K c − s 2 − To K o − s 2 (2.129)
ω c 2r1 Ic 2r1o = Tc K c − c 2r1 + Ti K o − c 2r1 − To (2.130)
Here,
I I a2 + I I (S / R )2 + I c1r 2 I s1 (R1S2 / S1 R2 )2
Is 2 = I s 2 + c 2r1o s1 2 c 2r1o c1r 2 2 2 (2.131)
I c 2r1o + I c1r 2 (1 + S2 / R2 )2 + I s1a12
64
6293_Book.indb 64 1/11/13 3:52 PM

I I (1 + S2 / R2 )2 + I c1r 2 I s1 (R1S2 / S1 R2 )2 + I s1 I s 2 a12

Ic 2r1o = I c 2r1o + c1r 2 s 2 (2.132)
I s 2 + I c1r 2 (S2 / R2 )2 + I s1a22
I c 2r 1o a2 + I c 1r 2a2 (R1 / R2 )
K c −s 2 = (2.133)
I c 2r 1o + I c 1r 2 (1 + S2 / R2 )2 + I s1a12
I c 1r 2a2 + I s1a1a2
K o −s 2 = (2.134)
I c 2r 1o + I c 1r 2 (1 + S2 / R2 )2 + I s1a12
I s 2a1 + I c 1r 2 (S22 R1 ) /(S1R22 )

K c −c 2r 1 = (2.135)
I s 2 + I c 1r 2 (S2 / R2 )2 + I s1a22
I c 1r 2a2 + I s1a1a2
K i −c 2r 1 = (2.136)
I s 2 + I c 1r 2 (S2 / R2 )2 + I s1a22
The coefficient Kc–c2r1 shown in Eq. 2.135 is the dynamic gain from the second clutch to the
output acceleration. Lower Kc–c2r1 means lower disturbance from the clutch to the output accel-
eration. Kc–c2r1 is a function not only of sun and ring gear size, but also is a function of inertias.
Therefore, adjusting inertias can minimize Kc–c2r1. However, keep in mind that we do not have
total flexibility in the selection of sun and ring gear size (selected for performance and fuel
economy) and inertia (function of packaging volume and driveline mass).
2.2.13 Six- and eight-speed planetary automatic

transmissions
Planetary automatic transmissions started from three and four speeds (gear ratio), and have
progressed to six, eight, and more speeds. This section presents examples of a six- and an
eight-speed planetary automatic transmission.
Figure 2.76 shows an example of six speeds. This six-speed transmission has three simple plan-
etary gear sets, and five clutches. The clutch state of the transmission is shown in Table 2.3. It
can be noticed that all single-step and double-step shifts are single-transition shifts (only one
oncoming and one off-going clutch). The Lagrange dynamic equation of this six-speed trans-
mission can be obtained using the guidelines given in Section 2.2.9, and is shown in Eq. 2.137.
MΩ = BT (2.137)
Fig. 2.76 Stick diagram of a six-speed

planetary automatic transmission.
65
6293_Book.indb 65 1/11/13 3:53 PM

66
Chapter 2
Here,
6293_Book.indb 66
 I s1 0 0 0 0 0 0 0 0 0 0 S1 0 0 0 0 0 0 0 
 
 0 Ir1 0 0 0 0 0 0 0 0 0 R1 0 0 0 0 −1 0 0 
 
 0 0 Ic1 0 0 0 0 0 0 0 0 −(S1 + R1) 0 0 1 0 0 0 0 
 0 0 0 Is2 0 0 0 0 0 0 0 0 S2 0 0 0 0 1 0 
 
 0 0 0 0 Ir 2 0 0 0 0 0 0 0 R2 0 −1 0 0 0 0 
 
 0 0 0 0 0 Ic 2 0 0 0 0 0 0 −(S 2 + R 2) 0 0 1 0 0 0 
 0 0 0 0 0 0 Is3 0 0 0 0 0 0 S3 0 0 0 0 0 
 
 0 0 0 0 0 0 0 Ir 3 0 0 0 0 0 R3 0 −1 0 0 0 
 0 0 0 0 0 0 0 0 Ic 3 0 0 0 0 −(S 3 + R 3) 0 0 1 0 1 
M = 
 0 0 0 0 0 0 0 0 0 I in 0 0 0 0 0 0 0 −1 0 
 
 0 0 0 0 0 0 0 0 0 0 Io 0 0 0 0 0 0 0 −1 
 S1 R1 −(S1 + R1) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
 
 0 0 0 S 2 R 2 −(S 2 + R 2) 0 0 0 0 0 0 0 0 0 0 0 0 0 
 0 0 0 0 0 0 S 3 R 3 −(S 3 + R 3) 0 0 0 0 0 0 0 0 0 0 
 
 0 0 1 0 −1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
 0 0 0 0 0 1 0 −1 0 0 0 0 0 0 0 0 0 0 0 
 
 0 −1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 
 0 0 0 1 0 0 0 0 0 −1 0 0 0 0 0 0 0 0 0 

 0 0 0 0 0 0 0 0 1 0 −1 0 0 0 0 0 0 0 0 
1/11/13 3:53 PM
6293_Book.indb 67
 ω s1 
 
 ω r1  
  0 0 1 1 0 0 0 
 
 ω c1 
 0 0 0 0 0 0 0 
 ω s 2  
  0 1 0 0 0 0 0 
 
 ω r 2   0 0 0 0 0 0 0 
   0 0 0 0 1 0 0 
 ω c 2 

 ω s 3  0 0 0 0 0 0 0   T 
   C 1234 
   1 0 0 0 0 0 0 
 ω r 3   TCB1R 
 0 0 0 0 0 0 0   
 ω c 3   
  0 0 0 0 0 0 0  TCB 26 
  
Ω = ω in  B = 0 0 0 −1 −1 1 0  T =  TC 35 R
   
 0 0 0 0 0 0 −1   TC 456 
 ω o 
 
 0 0 0 0 0  
F1   0 0 
 Tin 
   0 0 0 0 0 0 0 
 F2   To 
 0 0 0 0 0 0 0   
   
 F3 
 0 0 0 0 0 0 0 
 F4   0 0 0 0 0 0 0 
  
 F5  0 0 0 0 0 0 0 
 
 F6   0 0 0 0 0 0 0 
 
 F7 
 0 0 0 0 0 0 0 
 
 F8 
67
1/11/13 3:53 PM
Chapter 2
Figure 2.77 shows an example of an eight-speed planetary automatic transmission. It has four
simple planetary gear sets and five clutches. The clutch state of the eight-speed transmission is
shown in Table 2.4. It can be seen that all single-step and double-step shifts are single-transition
Table 2.3 Clutch State Table of the Six-speed Automatic Transmission

Gear C1234 CB1R CB26 C35R C456
1st gear X X
2nd gear X X
3rd gear X X
4th gear X X
5th gear X X
6th gear X X
Reverse gear X X
Fig. 2.77 Stick diagram of an eight-

speed planetary automatic transmission.
Table 2.4 Clutch State Table of the Eight-speed Automatic Transmission

Gear CB12345R CB1278R C13567 C23468 C45678R
1st gear X X X
2nd gear X X X
3rd gear X X X
4th gear X X X
5th gear X X X
6th gear X X X
7th gear X X X
8th gear X X X
Reverse gear X X X
68
6293_Book.indb 68 1/11/13 3:53 PM

shifts. The dynamic equation of the eight-speed transmission can be obtained in the same
manner as shown for the six speeds. However, the dimension of the matrices will increase.
The reader may already notice that the six-speed transmission has three open clutches when in
any of the six gears, and the eight-speed transmission has two open clutches when in a gear. This
means the eight-speed transmission has low power loss due to the drag force of open clutches.
Homework 2 for Section 2.2

1. Develop dynamic equations and a Simulink model of the transmission shown in
Fig. 2.78.
Fig. 2.78 Stick diagram of a four-speed transmission.

R1/S1 = 2.38, R2/S2 = 1.76
C12 C234 C13R C4 CR
1st gear X X
2nd gear X X
3rd gear X X
4th gear X X
Reverse gear X X
2. Build a generic dual-clutch Simulink model, and simulate the 1-2, 2-3, and 2-4 shift of the
transmission system in 1.
69
6293_Book.indb 69 1/11/13 3:53 PM

6293_Book.indb 70 1/11/13 3:53 PM

Chapter 3
Control of Planetary Gear
71
6293_Book.indb 71 1/11/13 3:53 PM

Chapter 3
A planetary gear automatic transmission is usually controlled by an electrohydraulic

control system and a microprocessor controller. The major control functionalities of
microprocessor controllers for automatic transmissions include: gear selection control,
gear shift control, line pressure control, torque converter clutch (TCC) control, and
failsafe control. In the following sections, we will start our discussions of controls of
planetary gear automatic transmissions from the discussion of the most important
control subsystem: the electrohydraulic pressure control system.
3.1 Electrohydraulic pressure control system

Let’s first introduce three basic equations that are indispensable for the dynamic modeling of
hydraulic control systems [3-1].
The first equation is the differential equation for describing the dynamic behavior of pressure
in a volume. Figure 3.1 illustrates the parameters of the system, and Eq. 3.1 is the differential
equation of the system.
β
p =
V
(Qi − Qo −V ) (3.1)
Here, p is the pressure in the volume, Qi is the flow into the volume, Qo is the flow out of the
volume, V is the volume value, and b is the bulk modulus of hydraulic fluid with a typical value
of 74,000,000 N/m2. The unit of in-and-out flow is m3/s, the unit of volume is m3, and the unit
of pressure is Pascal (N/m2).
The second equation is the flow equation of an orifice. Figure 3.2 illustrates the parameters of
the system, and Eq. 3.2 is the equation for calculating the orifice flow.
2
Q = sign ( p bef − paft )C d A p bef − paft (3.2)
ρ
Here, Q is the flow through the orifice, pbef is the pressure before the orifice, paft is the pressure
after the orifice, A is the area of the orifice, Cd is the unitless discharge coefficient with typical
value of 0.6, and r is the mass density of hydraulic fluid with a typical valve of 878 Kg/m3.
Fig. 3.1 Hydraulic fluid flow into and

out of a volume.
Fig. 3.2 Flow through an orifice.
72
6293_Book.indb 72 1/11/13 3:53 PM

Control of Planetary Gear Automatic Transmissions
The third equation is the pipe flow equation. The laminar flow through round pipes is given
in Eq. 3.3.
πr 4
Q= ( p1 − p2 ) (3.3)
8µ L
Here, Q is the flow rate, r is the radius of the pipe, L is the length of the pipe, p1 and p2 are the
terminal pressures, and m is the fluid viscosity.
3.1.1 Hydraulic pressure control system and its

simulation models
In automatic transmissions, the hydraulic line pressure, clutch pressure, and TCC pressure are
all controlled by electrohydraulic pressure control systems. A typical electrohydraulic pressure
control system consists of a proportional pressure control (PPC) solenoid and a spool-type
pressure regulation valve.
Let’s take a look at the PPC solenoid first, and discuss the spool-type pressure regulation valve
later. The cross-section view and descriptive drawing of a PPC solenoid are shown in Figs. 3.3a
and 3.3b, respectively.
The function of a PPC solenoid is to control the pressure in its control (output) pressure
chamber (see Fig. 3.3b). The hydraulic pressure in the control pressure chamber is controlled
by the electrical current of the coil assembly. Assume that the left end of the control pressure
Fig. 3.3a Cross section of a

proportional pressure control (PPC)
solenoid.
Fig. 3.3B Descriptive drawing

of a PPC solenoid.
73
6293_Book.indb 73 1/11/13 3:53 PM

Chapter 3
chamber is blocked. Then the pressure in the chamber is determined by the size of the feed
orifice, the supply pressure, and the opening of the variable orifice. The opening of the variable
orifice is determined by the position of the armature. The larger opening of the variable orifice
results in a lower pressure in the control pressure chamber, and vice versa. The hydraulic
pressure in the control pressure chamber presses on the armature in the opposing direction
of the electromagnetic force generated by the electric current in the coil assembly. This forms
a pressure feedback action. Due to this pressure feedback on the armature, the position of the
armature and, therefore, the opening of the variable orifice are automatically adjusted until
the force generated by the hydraulic pressure in the control chamber equals the electro
magnetic force. Therefore, changing the electric current will proportionally change the
hydraulic pressure.
The steady-state characteristics of a PPC solenoid can be described by the pressure-current
(P-I) curve, as shown in Fig. 3.4. The dynamic characteristics of a PPC solenoid can be
modeled as first-order lag, as shown in Eq. 3.4. The constant “a” in Eq. 3.4 defines the band-
width of the PPC solenoid, and p(i) is the P-I function shown in Fig. 3.4. The detailed dynamic
equation model of a PPC solenoid, based on the first principle of physics, will be given in a
later section.
P (s) a
= p(i ) (3.4)
I (s) s +a
Because the capability of passing flow is very limited due to their small feed orifice, PPC sole-
noids cannot be used to control clutches directly. The output pressure of a PPC solenoid is
usually fed to a spool-type pressure regulation valve as the control pressure. A pressure regula-
tion spool valve can carry high flow, and therefore is used to control clutches directly.
In a pressure regulation spool valve (see Fig. 3.5), the opening from the supply pressure port
to the output pressure port, and the opening from the output pressure port to the exhaust
port, are determined by the position of spool valve. The control pressure from a PPC solenoid
pushes on the area of As on the left end of the spool valve, and the output pressure pushes on
the area of Afb on the right end of the spool valve, through the feedback pressure passage and
the feedback orifice. Due to existence of this pressure feedback, the position of the spool valve
will be automatically adjusted until the net force acting on the spool valve equals zero. As a
result of the pressure feedback, the output pressure of a pressure regulation spool valve follows
the control pressure, with a steady-state gain of As/Afd.
Fig. 3.4 P-I curve of a

pressure control solenoid.
74
6293_Book.indb 74 1/11/13 3:53 PM

Fig. 3.5 Pressure

regulation spool valve.
The dynamic equations of a pressure regulation spool valve are given as follows. Equation 3.5
is the equation of motion of the spool. Equation 3.6 is the dynamic equation of the output
pressure. Equations 3.7–3.9 are flow equations. Equation 3.10 is the dynamic equation of the
feedback pressure [3-1].
In these equations, x is the spool position, pout is the output pressure, ps is the control pressure,
Psup is the supply pressure, pfb is the feedback pressure, As is the area of the control pressure
side, M is the mass of the spool, Afb is the area of the feedback pressure side, D is the damping
coefficient, k is the spring coefficient, xo is the spring preload, b is the buck modulus of the
hydraulic fluid, V is the volume of the output chamber, Cd is the discharge coefficient, Ain(x)
is the intake flow area as a function of the spool position x, Aout(x) is the outflow area as a
function of the spool position x, and r is the mass density of the hydraulic fluid. Ain(x) and
Aout(x) can be modeled using look-up tables, as shown in Fig. 3.13. Qin is the flow from the
supply pressure port to the output pressure port. Qout is the flow from the output pressure port
to the exhaust port. Qfd is the flow from the output port to the pressure feedback area. Qload is
the flow to the clutch piston connected to the output port.
xv = ( As ps − A fb p fb − Dv xv − kv (x v + x vo )) / M v ( x v < x limit ) (3.5)
β
p out =
Vload
(Qin − Qout − Q fb − Qload ) (3.6)
2
Qin = sign ( psup − pout )C d Ain ( x v ) psup − pout (3.7)
ρ
2
Qout = sign ( pout )C d Aout ( x v ) pout (3.8)
ρ
2
Q fb = sign ( pout − p fb )C d π rfbo
2
pout − p fb (3.9)
ρ
β
p fb =
V fb
(Q fb + Avb xv ) (3.10)
75
6293_Book.indb 75 1/11/13 3:53 PM

Chapter 3
In transmission hydraulic control schematics, components are represented by symbols. The

symbols for the PPC solenoids and pressure regulation spool valves are shown in Figs. 3.6 and
3.7, respectively.
Another important hydraulic control component used in automatic transmission controls is
the hydraulic piston. Hydraulic pistons are used to generate apply force on clutches. A hydrau-
lic piston system is depicted in Fig. 3.8.
The hydraulic fluid is fed into the piston chamber through an orifice with the pressure of pfeed.
The Pfeed usually comes from the output pressure port of a pressure regulation spool valve. The
hydraulic pressure pushes the piston leftward. The free travel h needs to be cleared before the
clutch-apply force could be established. The return spring pushes the piston back to the right-
end side.
Fig. 3.6 Symbol for proportional

pressure control solenoids.
Fig. 3.7 Symbols for

pressure regulation valves
controlled by a PPC solenoid.
Fig. 3.8 Diagram of a

hydraulic piston system.
76
6293_Book.indb 76 1/11/13 3:53 PM

Dynamic equations of a hydraulic piston system are given in Eqs. 3.11–3.13.

Equation 3.11 is the dynamic equation of the piston motion. Equation 3.12 is the dynamic
equation of the piston chamber pressure. Equation 3.13 is the flow equation of the clutch feed
orifice. Here, xp is the displacement of the piston, Ap is the area of the piston, pp is the hydraulic
pressure in the piston chamber, Pfeed is the feed pressure from a pressure regulation spool valve,
Mp is the mass of piston, D is the damping coefficient, k is the spring constant of the clutch
return spring, xo is the spring preload, Vo is the chamber volume when x = 0, and Aor is the area
of the orifice.
M p xp = A p p p − Dx − k ( x + xo ) (0 < x < h) (3.11)
β
p p = (Q − x A ) (3.12)
( o x p Ap ) load p p
V +
2
Q load = sign ( p pr − p p )C d Aor p feed − p p (3.13)
ρ
As a way to summarize the dynamic equations of hydraulic pressure and clutch control
systems, the Simulink models of a hydraulic clutch control system, consisting of a PPC
solenoid, a pressure regulation spool valve, and a hydraulic piston, are shown in Figs. 3.9–3.14.
Figure 3.9 is the top-level model, and the other figures are submodels. All those models are
based on the equations just given.
Figure 3.10 is the submodel of the pressure regulation spool valve based on Eqs. 3.5–3.10.
Figure 3.11 is the submodel of the hydraulic piston based on Eqs. 3.11–3.13. Figure 3.12 is the
submodel of the orifice flow calculation that is used to implement flow equations (Eqs. 3.7–3.9
and 3.13). Figure 3.13 is the submodel of the spool valve opening calculation, which uses two
look-up tables to implement the flow-area function, Ain(x) and Aout(x), which is used in Eqs.
3.7 and 3.8. Figure 3.14 is the submodel of the spring-mass motion system that is used for both
clutch piston motion (Eq. 3.5) and spool valve motion (Eq. 3.11).
Fig. 3.9 Simulink model of the clutch control system (top-level model).
77
6293_Book.indb 77 1/11/13 3:53 PM

Chapter 3
Fig. 3.10 The submodel of the pressure regulation valve.
Fig. 3.11 The submodel of the piston.
78
6293_Book.indb 78 1/11/13 3:53 PM

Fig. 3.12 The submodel of the orifice flow calculation.
Fig. 3.13 The submodel of the spool valve opening calculation.
Fig. 3.14 The submodel of clutch piston motion or the spool valve motion.
79
6293_Book.indb 79 1/11/13 3:53 PM

Chapter 3
3.1.2 Detailed model of PPC solenoids

In the early section, we modeled a PPC solenoid as a first-order system with a P-I curve.
In this section, let’s consider the detailed model of it. Refer to Fig. 3.3b for a description of
PPC solenoids.
The first part of the model of a PPC solenoid is the electric circuit model. The input of the
model is the voltage applied to the coil; the output is the electric current in the coil. The
dynamic equation of the electric circuit is shown in Eq. 3.14.
i = (ei − Ri − E v ixa ) / L (3.14)
Here, ei is the voltage applied to the coil, i is the electric current in the coil, R is the electric
resistance, L is the inductance, Ev is the back emf coefficient, and xa is the displacement of the
armature.
A closed-loop control can be used to control electric current in the coil. Since the controlled
object is a first-order system, a high-gain proportional feedback control is sufficient. The
equation of the current feedback control is shown in Eq. 3.15.
ei = k p (ir − i ) (3.15)
Here, ir is the desired current, and kp is the proportional gain.

The second part of the model is the dynamic motion equation of the armature; it is shown in
Eq. 3.16.
M a xa = E f i 2 − Aafb paout − Da xa − ka ( x a + x ao ) ( xa < x limit ) (3.16)
Here, Ma is the mass of the armature, Ef is the magnetic force coefficient, Aafb is the area of
pressure feedback, Paout is the output pressure, Da is the damping coefficient, ka is the spring
constant, and xao is the spring preload.
The last part of the model is the dynamic equation of output pressure paout; it is shown in
Eq. 3.17 (see also Eqs. 3.18 and 3.19).
β
paout = (Qain − Qaout ) (3.17)
Va
2
Qain = sign ( psup − paout )C d Aain psup − paout (3.18)
ρ
2
Qaout = sign ( pout )C d Aaout ( x a ) paout (3.19)
ρ
Here, Aain is the flow area of the supply orifice, Aaout(xa) is the flow area of the variable orifice as
a function of xa, and Va is the volume of the control pressure chamber.
80
6293_Book.indb 80 1/11/13 3:53 PM

3.1.3 High-flow, direct-acting pressure control valves

The pressure control system described in the previous sections is a two-stage system; it uses
a low-flow pressure control solenoid to control a high-flow pressure regulation valve. The
alternative pressure control system is a single-stage system, in which the high-flow pressure
regulation valve is directly driven by an electromagnetic force motor. The conceptual diagram
of such a pressure control system is shown in Fig. 3.15. Dynamic equations of such a high-flow,
direct-action pressure control valve can be obtained by combining Eqs. 3.14 and 3.15 with the
dynamic equations of the pressure regulation valve developed previously. The result is shown
in Eqs. 3.20–3.27.
i = (ei − Ri − E v ixv ) / L (3.20)
ei = k p (ir − i ) (3.21)
xv = ( As ps − A fb p fb − Dv xv − kv ( x v + x vo )) / M v ( xv < x limit ) (3.22)
β
p out =
Vload
2
ρ
2
ρ
2
2
pout − p fb (3.26)
ρ
β
p fb =
V fb
(Q fb + Avb xv ) (3.27)
Fig. 3.15 High-flow, direct-

acting pressure control valve.
81
6293_Book.indb 81 1/11/13 3:53 PM

Chapter 3
3.1.4 Pulse width modulated (PWM) solenoid

Another type of pressure control device used in automatic transmissions is the pulse width
modulated (PWM) solenoid type shown in Figs. 3.16 and 3.17. The electromagnetic force
generated by the coil assembly and the spring force move the metering ball (armature) left and
right; they alternatively open the output pressure port to the supply pressure port or exhaust
port. The supply pressure port is supplied by a constant pressure source.
Unlike a PPC solenoid, in a PWM solenoid there is no output pressure feedback on the
armature (metering ball). Therefore, the output pressure port will be either fully open to the
supply pressure, or fully open to the exhaust. The level of output pressure is controlled by con-
trolling the period in which the output pressure port opens to the supply pressure port in one
PWM cycle. Figure 3.18 shows the input PWM signal, armature position, and output pressure
of a PWM solenoid. The percentage of the time the output port is open to the supply pressure
port in one cycle is called the duty cycle. The average of the output pressure is determined by
the duty cycle of the input signal and supply pressure. The case shown in the left chart of Fig.
3.18 is the case of a 30% duty cycle. The right chart is for a 70% duty cycle. Higher duty cycle
results in higher output pressure.
The output pressure of a PWM solenoid has a pressure ripple, as shown in Fig. 3.18. The mag-
nitude of the output pressure ripple can be reduced by increasing the hydraulic capacitance of
the output load, characterized by the value of DV/Dp. Here, DV is the volume change and Dp is
the pressure change. Sometimes an accumulator, or a hydraulic capacitor, shown in Fig. 3.19, is
necessary to reduce the pressure ripple to a desired level.
Fig. 3.16 Cross-section view of a pulse width modulated (PWM) solenoid.
Fig. 3.17 Descriptive

drawing of a pulse width
modulated (PWM) solenoid.
82
6293_Book.indb 82 1/11/13 3:53 PM

Fig. 3.18 Input signal, armature position, and output pressure of a PWM solenoid.
Fig. 3.19 Diagram of hydraulic

capacity (accumulator).
Next, let’s consider the dynamic mathematical model of a PWM solenoid. To simplify the
model, let’s assume that the input to the model is a pulse width modulated position command
of the metering ball. Then the physical system under consideration becomes as shown in Fig.
3.20. The output port of the PWM solenoid is connected to a ripple reduction accumulator.
The position of the metering ball is the output signal X of the first-order lag. The input signal
to the first-order lag is a pulse width modulated position command, Xpwm. The position of the
metering ball ranges from 0 to 1. When it is 0, the output port is fully open to the exhaust port,
and closed to the supply port. When it is 1, the output port is closed to the exhaust port, and
fully open to the supply port.
Fig. 3.20 PWM pressure control system.
83
6293_Book.indb 83 1/11/13 3:53 PM

Chapter 3
The dynamic equations of this PWM pressure control system are given in Eqs. 3.28–3.32. Here,
po is the output pressure of the PWM solenoid valve, ps is the supply pressure that is constant,
Ao is the diameter of the accumulator feed orifice, pa is the accumulator pressure, Ap is the area
of the accumulator piston, k is the stiffness coefficient of the accumulator spring, Ain is the
diameter of the supply port of the PWM solenoid valve, and Aout is the diameter of the exhaust
port. Equation 3.28 is the dynamic equation of the PWM valve output pressure. Equation 3.32
is the dynamic equation of the accumulator pressure. Equation 3.33 is the dynamic motion
equation of the accumulator piston. As described by Eqs. 3.29 and 3.30, when the position of
the metering ball X is 1, the in-flow Qin is non-zero, and the out-flow Qout is zero. When X is 0,
Qin is zero, and Qout is non-zero.
β
po = (Qin − Qout − Qorif ) (3.28)
Vo
2
Qin = sign ( ps − po )C d Ain X ps − po (3.29)
ρ
2
Qout = sign ( po )C d Aout (1 − X ) po (3.30)
ρ
2
Qorif = sign ( po − pa )C d Ao po − pa (3.31)
ρ
β
pa = (Q − x A ) (3.32)
( o x p Ap ) orif p p
V +
M p xp = A p pa − Dx p − k ( x p + xo ) (0 < x < h) (3.33)
3.1.5 Analytical study of hydraulic clutch control systems

A hydraulic clutch control system is a system with the command to the pressure control
solenoid as the input, and the clutch-apply force as the output. It is the most important sub-
control system in clutch-to-clutch shift automatic transmissions for achieving superior gear
shift quality. Equipped with the knowledge of dynamic modeling of hydraulic control systems
presented in the last section, this section focuses on the analytical study of hydraulic clutch
control systems. The purpose of this analytical study is to introduce some necessary analysis
methods, to illustrate key system design issues, and to provide guidelines for achieving optimal
design of the system.
A hydraulic clutch control system typically consists of a pressure control solenoid, a spool-
type pressure regulation valve, a hydraulic piston, and various accessory elements to help form
a desired static and dynamic characteristic of the system. A hydraulic clutch control system
performs two major control tasks. One is clutch fill control, and the other is clutch-apply-
force control. The objective of clutch fill control is to move the clutch piston forward as fast
as possible to achieve contact with the clutch plates. The major difficulty of clutch fill control
84
6293_Book.indb 84 1/11/13 3:53 PM

is to prevent apply-force surge at the end of piston stroke. The objective of clutch-apply-force
control is to ensure that apply force follows the command with the desired response speed,
precision, and stability. Therefore, the response time, control precision, and robustness are
important design criteria.
Analytic studies of a hydraulic control system usually start from establishing dynamic equa-
tions of the system. Then block diagrams and transfer functions of the system are derived from
dynamic equations. Block diagrams and transfer functions are used to analyze the effect of
system parameters to system performance by using analysis methods such as Nyquist, Bode,
and Root Locus Plots. Dynamic equations are also used to create computer dynamic simula-
tion models of the system. Such simulation models are widely used in analysis and design of
hydraulic control systems.
In the following discussions, to make this section independently readable, some contents in
the previous sections are repeated.
3.1.5.1 Dynamic equations

Establishing dynamic equations is the starting point of analytical study of hydraulic clutch
control systems. This section uses two examples to introduce the basic dynamic equations for
analytical study of hydraulic clutch control systems. These dynamic equations can be used to
derive system transfer functions or to create computer simulation models.
In the first example, let’s consider a spool-type pressure regulation valve, as shown in Fig. 3.21.
The dynamic equations of the system have been given in Section 3.1.1, and for the reader’s
convenience they are repeated in Eqs. 3.34–3.39. Here, psup is the supply pressure, ps is the
signal pressure, pout is the output pressure, pfb is the feedback pressure, As is the area for signal
pressure ps to push on, Afd is the area for feedback pressure pfb to push on, kv is the spring
constant, rfbo is the radius of the feedback orifice, Mv is the mass of the spool valve, xv is the
displacement of the spool valve, and Ain(xv) and Aout(xv) are the flow areas of in and out flows
of the valve as functions of valve displacement xv. Equation 3.34 is the motion equation of the
spool; Eq. 3.35 is the differential equation of the output pressure; Eqs. 3.36–3.38 are flow equa-
tions of the in-port, out-port, and feedback path, respectively; and Eq. 3.39 is the differential
equation of the feedback pressure.
xv = ( As ps − A fb p fb − Dv xv − kv ( x v + x vo )) / M v ( xv < x limit ) (3.34)
Fig. 3.21 Spool-type pressure regulation valve.
85
6293_Book.indb 85 1/11/13 3:53 PM

Chapter 3
β
p out =
Vload
2
ρ
2
ρ
2
2
pout − p fb (3.38)
ρ
β
p fb =
V fb
(Q fb + Avb xv ) (3.39)
As the second example, let’s consider a hydraulic piston system as shown in Fig. 3.22. The
system is slightly different from the piston system given in Section 3.1.1. The system includes
a wave plate, which is a compliance element between piston and clutch plates. The dynamic
equations of the system are given in Eqs. 3.40–3.42. Here, rclfd is the radius of the clutch feed
orifice, pp is the pressure pushing on the piston, Ap is the piston area, Fdrag(pp) is the piston seal
drag force that is a function of pp, Dp is the coefficient of damping, kp is the coefficient of the
return spring, xp is the displacement of the piston, Mp is the mass of piston, pout is the output
pressure of the pressure regulation valve, Fapl is the clutch-apply force (which is a function
of xp, with a typical shape as shown in Fig. 3.23). In Fig. 3.23, h1 is the clutch clearance, and
h2 is the wave plate travel. The force curve between the points of h1 and h1 + h2 is the wave
plate force curve; the force curve after h1 + h2 is very stiff, which represents the mechanical
compliance of the compressed clutch plates. Note that the clutch-apply force is a function of
deformation of the wave plate until the wave plate is solidly compressed. Then the apply force
can be determined directly by the pressure pp. Equation 3.40 is the dynamic motion equation
of the piston, Eq. 3.41 is the differential equation of the piston pressure, and Eq. 3.42 is the
clutch-apply-force function that is shown in Fig. 3.23.
Fig. 3.22 Hydraulic piston

system with a wave plate.
86
6293_Book.indb 86 1/11/13 3:53 PM

Fig. 3.23 Clutch-apply-force

function Fapl = f(xp).
xp = { A p p p − D p x p − Fdrag ( p p ) tanh ( x p / α ) − k p ( x p + x po ) − Fapl } / M p (3.40)
β  2 
p p = sign ( pout − p p )C d π rclfd
2
pout − p p − x p A p  (3.41)
(Vo + x p Ap )  ρ 
Fapl = f ( x p ) (3.42)
3.1.5.2 Block diagrams and transfer functions

Block diagrams and transfer functions are widely used in control system analysis. Block
diagrams and transfer functions of various hydraulic clutch control systems can be derived
from their dynamic equations. This section presents block diagrams and transfer func-
tions of two pressure regulating systems as examples. Readers are encouraged to view the
reference [3-1].
The first example is a spool-type pressure regulation valve with a dead-end output port, as
shown in Fig. 3.24.
The block diagram of the system is shown in Fig. 3.25. Here, the transfer function (TF) of
the spool is given in its damping ratio and natural frequency form. The damping ratio x and
natural frequency wn are defined in Eqs. 3.43 and 3.44. Function Ain(xv) in the block diagram
provides the relationship between the displacement of spool xv and the hydraulic flow area
from supply pressure port to output pressure port. Function Aout(xv) provides the hydraulic
flow area from output pressure port to exhaust port. These flow-area functions are determined
by the geometric design of the spool valve and ports.
Fig. 3.24 PR valve with a

dead-end output port.
87
6293_Book.indb 87 1/11/13 3:53 PM

Chapter 3
Dv
ξ= (3.43)
2 kv M v
kv
ωn = (3.44)
Mv
The block diagram in Fig. 3.25 has some nonlinear blocks. To prepare for future discussions,
these nonlinear blocks are linearized using the local linearization method. The resulting block
diagram is shown in Fig. 3.26. The flow gains, kqin, kqout, and pressure-flow coefficients, kpin,
kpout, obtained from linearization, are given in Eqs. 3.45–3.48. The linearized block diagram
can be used to perform stability and response analysis using methods such as Nyquist, Bode,
and Root Locus Plots. In Eqs. 3.45–3.48, psupo, pouto, pvo, and plo are the points where local lin-
earization is performed.
2 dAin ( x v )
kqin = C d psupo − pouto (3.45)
ρ dx v xv =xvo
2 dAout ( x v )
kqout = C d pouto (3.46)
ρ dx v x v = x vo
C d Ain ( x vo )
k pin = (3.47)
2 ρ psupo − pouto
C d Aout ( x vo )
k pout = (3.48)
2 ρ pouto
The second example is a spool-type pressure regulation valve with a dead-end output port and
a supply-side orifice, as shown in Fig. 3.27. The nonlinear and linearized block diagrams are
shown in Figs. 3.28 and 3.29, respectively.
Fig. 3.25 Block diagram of pressure regulation system.
88
6293_Book.indb 88 1/11/13 3:53 PM

Fig. 3.26 Linearized block diagram of pressure regulation system.
Once again, the linearized block diagram in Fig. 3.29 can be used to perform a stability and
response analysis using the frequency response method and root locus method. Those classical
analyses can provide significant insights into the characteristics of the pressure control system.
The classical control system analysis techniques can be supplemented with dynamic simulation
techniques. Dynamic simulation approaches also allow the control systems to interact with
control algorithms.
Fig. 3.27 PR valve with

dead-end output port and a
supply orifice.
Fig. 3.28 Block diagram of pressure regulation system with a supply orifice.
89
6293_Book.indb 89 1/11/13 3:53 PM

Chapter 3
Fig. 3.29 Linearized block diagram of pressure regulation system with a supply orifice.
Fig. 3.30 Simulink model of a pressure control system.
90
6293_Book.indb 90 1/11/13 3:53 PM

3.1.5.3 Dynamic simulation models

Dynamic simulation is another control system analysis method. Dynamic simulations are per-
formed by numerically solving dynamic equations of the system with a computer. There are
various computer software products available for this purpose. One popular one is Simulink.
An example of a Simulink model of a pressure control system is shown in Fig. 3.30. The model
is built based on dynamic equations of the system, which are presented in Eqs. 3.34–3.39.
Simulink requires users to define equations of the system to build simulation models; some
other simulation software packages (such as AMESim) do not require users to define system
equations. An example of an AMESim model of a hydraulic clutch control system is shown
in Fig. 3.31. The simulation model is built with components that come with the AMESim
software. There are mathematical equations behind each of those components.
In this AMESim model, seal drag force and wave plate force are modeled with user-
supplied functions.
3.1.5.4 Discussion of hydraulic clutch control system design

Using the dynamic analysis methods described above, this section will discuss some design
issues of hydraulic clutch control systems.
The first discussion is about the design of the spool valve. The spool valve is the most impor-
tant component in hydraulic clutch control systems. The major design parameters of a spool
valve include spring rate, feedback orifice, mass and flow area characteristics that are deter-
mined by the diameter of spool valve, degree of overlap or under-lap, and detailed geometry of
the flow control area.
Spring rate, feedback orifice, and mass determine the motion behavior of the spool. The influ-
ence of those three parameters can be analyzed using the transfer function (TF) of the spool
shown in the block diagram of Fig. 3.25. For convenience, the transfer function is repeated in
Eq. 3.49 (see also Eqs. 3.50 and 3.51).
Fig. 3.31 AMESim model of a hydraulic clutch control system.
91
6293_Book.indb 91 1/11/13 3:53 PM

Chapter 3
The damping ratio coefficient x should be near 1 to provide adequate damping effect. An
over-damped system (x>>1) will have a slow response. An under-damped system (x<<1) will
be unstable. If x = 1, the time constant of the system will be 1/wn. This means that a larger wn
provides a faster response. The steady-state gain of the system is 1/kv. The system damping
characterized by the damping coefficient Dv is mainly determined by the size of the feedback
orifice. A smaller feedback orifice means a high damping coefficient. The spring rate kv, mass
Mv, and linear damping Dv are determined by many design tradeoff considerations. Their
influences to the system dynamic characteristics are summarized here.
• A lower Mv provides a larger wn. Therefore, the mass of the spool should be as low as
possible. Attaching additional control functionalities to a spool valve will increase its
mass, and therefore should be avoided.
• A higher kv provides a larger wn, and a lower steady-state gain. Therefore, tradeoffs must
be made.
• The feedback orifice could be used to achieve an adequate damping ratio x.
x v (s) ωn2
= (3.49)
e(s) k (s 2 + 2ξωn + ωn2 )
Dv
ξ= (3.50)
2 kv M v
kv
ωn = (3.51)
Mv
The flow area characteristic of a spool valve is described by flow-area functions Ain(xv) and
Aout(xv), and is the single most important design parameter. Its importance can be observed
from the block diagram of Fig. 3.26. Flow gains kqin and kqout, which have expressions shown
in Eqs. 3.45 and 3.46, are major contributors to the open-loop gain of the system; therefore,
they are the major contributors to the stability and dynamic response of the system. Figures
3.32 and 3.33 present two different types of flow-area functions. Figure 3.32 shows flow-area
functions of a spool valve with a sharp-edged corner spool and a large overlap. These flow-area
functions have a dead band, which will degrade control performance. In contrast, Fig. 3.33
shows flow-area functions of a spool valve with a notched spool and a small overlap. The flow-
area functions become much smoother, and the dead band is eliminated. Valves with this type
of flow-area function will perform much better. Figures 3.34 and 3.35 compare step responses
Fig. 3.32 Flow-area functions of a spool valve with annular orifice, sharp-edged corner spool, and large overlap.
92
6293_Book.indb 92 1/11/13 3:53 PM

Fig. 3.33 Flow-area functions of a spool valve with a notched valve (or notched port) and small overlap.
Fig. 3.34 Simulation results of step responses of a PR system with a small overlap.
Fig. 3.35 Simulation results of step responses of a PR system with a large overlap.
93
6293_Book.indb 93 1/11/13 3:53 PM

Chapter 3
of systems with small overlap and large overlap, respectively. It can be seen that large overlap
degrades system performance significantly, because the spool has to travel a longer distance to
regulate pressure.
The second discussion is about two different concepts of pressure regulation valves for clutch
controls. The difference is in the way the pressure regulation (PR) valve is used during clutch
fill controls.
The first concept uses the PR valve as a trim valve during the clutch fill control. The system
configuration of such a concept is shown in Fig. 3.36. The major identifiers of such systems
are: (1) the supply pressure side orifice, and (2) the wave plate in the clutch pack. The idea is
that during the clutch fill phase, although the commanded pressure will be kept just somewhat
above the clutch return spring pressure, due to the existence of the supply-side orifice, the
spool valve will be at the fully open position. At the end of the piston stroke, the output
pressure will increase, and through the feedback pressure, it pushes the spool valve toward the
regulating position. However, due to the reaction delay of the spool valve, pressure overshoot-
ing could happen. This is why the system has a wave plate (or an accumulator) to provide the
needed additional compliance to reduce pressure overshoot. The pressure command, spool
valve response, and output pressure response of such a system are illustrated in Fig. 3.37.
The supply-side orifice and wave plate have some negative impacts on control performances of
the system. This will be discussed in the following sections.
Fig. 3.36 Concept of

trimming clutch fill control.
Fig. 3.37 Pressure command,

spool position, and output pressure
response of fill trim control system.
94
6293_Book.indb 94 1/11/13 3:53 PM

The second concept is the regulated clutch fill control, as shown in Fig. 3.38. Instead of the
supply-side orifice, a clutch feed orifice is placed between the PR valve and the clutch piston.
The pressure command, spool position, and clutch pressure responses of such a system are
shown in Fig. 3.39. During the clutch fill phase, the output pressure of the PR valve is com-
manded high to rapidly move the clutch piston. Near the end of the piston stroke, the pressure
command is dropped to a lower level to prevent pressure overshoots. An extra compliance
element is not needed in such a system. The fill time shown in Fig. 3.39 is the key control
parameter to achieve overshoot-free fast piston stroking. An adaptive learning algorithm is
usually used to control the clutch fill time, and will be discussed in a later section. The
purpose of the clutch feed orifice is to increase consistency of clutch fill time. This is because
the clutch feed pressure before the clutch feed orifice is regulated, and the clutch feed orifice is
a sharp-edged orifice whose flow characteristics are less sensitive to viscosity change of trans-
mission fluid.
After the discussions in the following two sections, it will be clear that the second concept of
clutch control configuration provides better control performance.
The third discussion is about the supply-side orifice. In Section 3.1.5.1, the block diagrams of
pressure regulation valves with and without supply-side orifice are derived. Now let’s consider
the effects of supply orifice to the system performance. Comparing the block diagram of the
system without supply orifice shown in Fig. 3.26 with the block diagram of the system with
Fig. 3.38 Concept of regulated clutch

fill control.
Fig. 3.39 Pressure command, spool position,

output pressure, and clutch pressure response
of regulated fill control system.
95
6293_Book.indb 95 1/11/13 3:53 PM

Chapter 3
supply orifice shown in Fig. 3.29, it can be seen that the supply orifice creates an additional
feedback loop. This additional feedback loop could make the system less stable. One can learn
this by deriving closed-loop transfer functions of these two block diagrams. We will demon-
strate these using simulation techniques in the next section.
The fourth discussion is about wave plates. Wave plates can be used to provide additional
compliance to help eliminate apply-force surge at the end of a clutch stroke. However, wave
plates with heavy loads and long strokes can have negative impacts to system performances.
For example, if the wave plate is not at a fully compressed state, the clutch-apply force will be
determined by the displacement of the clutch piston, not the hydraulic pressure. The presence
of the seal drag force does not allow a steady-state, one-to-one corresponding relation-
ship between the hydraulic pressure and piston displacement. This creates a huge hysteresis
between hydraulic pressure and clutch-apply force, and will make clutch-apply-force control
inconsistent.
For power-on down-shifts, it is highly desirable to be able to bring the oncoming clutch-apply
force rapidly from near zero to its full capacity in a controlled manner. However, a heavy wave
plate will make this difficult.
To demonstrate these arguments and also arguments in the previous section, dynamic simula-
tions are used with the simulation model shown in Fig. 3.31. A supply orifice is placed in the
model to demonstrate the arguments made in Section 3.1.5.4 about influence of the supply
orifice. First, the model is calibrated to demonstrate the effect of the supply orifice and wave
plate by using a 1.7-mm-diameter supply orifice, a wave plate curve as shown in Fig. 3.40,
and a large clutch feed orifice of 10 mm. This system configuration is the trimmed clutch fill
control system. Second, to show the result of removing the supply orifice and wave plate, the
diameter of the supply orifice is changed to 10 mm, the wave plate curve is changed to the
Fig. 3.40 Force-displacement

curve of a heavy wave plate.
Fig. 3.41 Force-displacement c

urve of a light wave plate.
96
6293_Book.indb 96 1/11/13 3:53 PM

one shown in Fig. 3.41, and the diameter of the clutch feed orifice is change to 3 mm. This
system configuration is the regulated clutch fill control system. The drag force curves used in
both simulations are the same as shown in Fig. 3.42. All other model parameters are also held
constant. The simulation results are shown in Figs. 3.43–3.46.
Figure 3.42 shows the response to small step commands of the system with a supply orifice and
a heavy wave plate. Because the wave plate has a long travel distance and a heavy load, it has
not reached its fully compressed state yet. Although the clutch pressure follows command, due
to the existence of seal drag force, the piston does not move accordingly; therefore, the clutch-
apply force does not follow the pressure command. The piston is pushed overboard by the
initial pressure overshoot, and held by seal drag force. Pressure fluctuations shake the piston
“loose” to some degree and result in some backstrokes. The seal drag force creates hysteresis
and causes such behaviors. Therefore, the magnitude of the seal drag force is a determining
factor. Lower seal drag force will present less impact on the controllability. Also, in this system
the pressure response exhibits overshoot and oscillation; this is due to the existence of the
supply-side orifice.
Figure 3.44 is the response to small step commands of the system without a supply-side orifice
and with a light wave plate. Because the wave plate is light, for the same level of command,
the wave plate is fully compressed; therefore, the clutch pressure directly determines the
clutch-apply force. The clutch pressure and apply force follows commands well and exhibits
no overshoot. Despite less compliance, there is no pressure overshoot and oscillation; this
indicates that the elimination of pressure overshoot and oscillation is due to the removal of the
supply orifice. In other words, the supply orifice does have a negative influence on the system
stability, as discussed in Section 3.1.5.4.
Figure 3.45 is the response to a large step command of the system with a supply orifice and a
heavy wave plate. The long travel of the wave plate limits the system response speed.
Figure 3.46 is the response to a large step command of the system without a supply orifice and
with a light wave plate. The system response is much faster.
The simulation results shown in Figs. 3.47–3.49 reveal another problem of the system with
a supply orifice and a heavy wave plate. In these three simulations, the pressure command
increases to the same level at different ramp-up rates. Figure 3.47 shows ramp up at a slow
rate, Fig. 3.48 illustrates a little faster ramp-up rate, and Fig. 3.49 is a step up to the same level.
Although the clutch pressure of three simulations reaches the same steady-state level, the
clutch-apply force ends at different levels. The cause of this is the compound effect of pressure
overshooting, seal drag force, and wave plate.
From the previous simulation results, it can be concluded that the supply-side orifice and
heavy wave plate have negative impacts on the system controllability and stability. The regu-
lated clutch fill control system performs better than the trimmed clutch fill control system.
Fig. 3.42 Seal drag force curve.
97
6293_Book.indb 97 1/11/13 3:53 PM

Chapter 3
As mentioned previously, the fill time is the key control parameter of the regulated clutch fill
control system. Fill-time learning algorithms, which have been developed to improve the pre-
cision of clutch fill controls, will be discussed in a later section.
Fig. 3.43 Simulation result of the response to a small step command of the system with a supply orifice and a heavy wave
plate.
98
6293_Book.indb 98 1/11/13 3:53 PM

Fig. 3.44 Simulation result of the response to a small step command of the system without a supply orifice, and with a
light wave plate.
99
6293_Book.indb 99 1/11/13 3:53 PM

Chapter 3
Fig. 3.45 Simulation result of the response to a large step command of the system with a supply orifice and a heavy wave
plate.
100
6293_Book.indb 100 1/11/13 3:53 PM

Fig. 3.46 Simulation result of the response to a large step command of the system without a supply orifice, and with a
light wave plate.
101
6293_Book.indb 101 1/11/13 3:53 PM

Chapter 3
Fig. 3.47 Simulation result of slow ramp-up response of the system with a supply orifice and a heavy wave plate.
102
6293_Book.indb 102 1/11/13 3:53 PM

Fig. 3.48 Simulation result of faster ramp-up response of the system with a supply orifice and a heavy wave plate.
103
6293_Book.indb 103 1/11/13 3:53 PM

Chapter 3
Fig. 3.49 Simulation result of step-up response of the system with a supply orifice and a heavy wave plate.
104
6293_Book.indb 104 1/11/13 3:53 PM

Homework for Section 3.1

1. Develop a Simulink model for the hydraulic clutch control system shown in Fig.
3.50. Here, Cd = 0.6, r = 878 Kg/m3, b = 74000000 N/m2, the piston area is 0.008
m2, h = 0.003 m, the return spring constant is 180000 N/m, the spring preload is
0.003 m. The diameter of the orifice is 0.0025 m.
2. Discuss the control methodology of fast-forwarding the piston to clear the
clearance h, and then bring the cylinder pressure to a target pressure using the
Simulink model developed in 1.
3. Develop a Simulink model of the PWM solenoid shown in Section 3.1.4, and
discuss the design of the hydraulic accumulator.
Fig. 3.50 Diagram of a

hydraulic clutch control system.
3.2 Clutch-to-clutch gear-shift control

To shift from one gear to another in an automatic transmission, one clutch needs to be
released, and another clutch needs to be applied. The control of such clutch-to-clutch shifts is
conducted based on an understanding of the fundamentals of gear-shifting mechanics. The
major objective of clutch-to-clutch shift controls is to minimize the output torque disturbance
while being able to complete the shift in the desired time duration. In this section, we first
discuss the gear-shifting mechanics from a control perspective, and then focus our attention
on controls.
3.2.1 Gear-shifting mechanics from a control perspective

In an early section we discussed gear-shifting mechanics from a static analysis perspective; in
this section we will have more discussions from a control perspective. Some contents discussed
in the early section will be repeated to make this section independently readable.
First, let’s define the terms “clutch torque capacity” and “clutch torque.” “Clutch torque
capacity” is the maximum torque a clutch can carry; it is determined by the clutch-apply force,
and can be expressed as F · m · R · n. Here, F is the clutch-apply force, m is the friction coeffi-
cient, R is the equivalent clutch radius, and n is the number of friction faces. The term “clutch
torque” means the torque actually carried by the clutch; it is determined by the external load
when the slip speed of the clutch is zero, and will equal clutch torque capacity when the slip
speed of the clutch is non-zero.
105
6293_Book.indb 105 1/11/13 3:53 PM

Chapter 3
Figure 3.51 shows the lever diagram of a two-speed planetary gear train. The letters a, b, and
c indicate the length of each section of the lever. When the clutch C1 is applied and clutch
C2 is released, the system is in first gear. When the clutch C2 is applied and clutch C1 is
released, the system is in second gear. Tn1 and Tn2 are torques exerted on nodes N1 and N2 by
clutches C1 and C2, respectively. Tin is the input torque. Tout is the output torque. Line L1 is the
level position of first gear. Line L2 is the level position of second gear. Line L0 is the zero
speed position.
In the first gear, clutch C1 grounds the node N1. So the speed of node N1 is zero. The vertical
distances from nodes on line L1 to the corresponding nodes on the zero speed line represent
the speeds of those nodes when the system is in first gear. The distance below the horizontal
line L0 indicates positive speed. The distance above the horizontal line L0 indicates negative
speed. Therefore, when in first gear, node N2 has a negative speed; thus, clutch C2 has a
negative slip speed.
In second gear, clutch C2 grounds node N2. So N2 has zero speed. The vertical distances from
nodes on line L2 to the corresponding nodes on line L0 represent the speeds of those nodes
when the system is in second gear. When shifting from first gear to second gear, the speed of
node N1 will be accelerated, and the speed of the input node will be decelerated. The decelera-
tion of the input should be recognized as the familiar drop in input speed that a driver will
see on the tachometer during an up-shift. The output speeds before and after gear shifting are
assumed to remain unchanged due to the large vehicle mass.
There are four types of shifts between first gear and second gear. Those are: power-on up-shift
(1-2), power-on down-shift (2-1), power-off up-shift, and power-off down-shift. When the
input torque is high enough to accelerate input speed, the shift is called power-on shift; oth-
erwise, the shift is called power-off shift. In the following, the focus will be on power-on
up-shifts and down-shifts. With an understanding of mechanics of these shifts, other types of
shifts are self-explanatory.
First let’s consider power-on up-shift. The power-on up-shift from first gear to second gear
is divided into two phases. The first phase is the torque phase. This is the phase in which the
torque will be transferred from clutch C1 to clutch C2. In the torque phase, the capacity of
the oncoming clutch C2 is increased from zero at a predetermined rate (see Fig. 3.52). As the
capacity of C2 (oncoming clutch) increases, the torque of C1 (off-going clutch) decreases. This
is because the torques Tin, Tn1, and Tn2 have the relationship shown in Eq. 3.52. Equation 3.52
is obtained by applying the principle of balance of moment about the output node. From the
equation it can be seen that because the input torque Tin is assumed to be constant during the
torque phase, as the torque of clutch C2, Tn2, increases, the torque of C1, Tn1, will decrease to
keep Eq. 3.52 satisfied. Then during the torque phase, as the torque capacity of C2 is increased,
what will happen to the output torque? This could be answered by Eq. 3.54. Equation 3.54 is
obtained by applying the principle of balance of moment about the node N1, or by substitut-
ing Eq. 3.52 into Eq. 3.53. Equation 3.54 shows that as Tn2 increases, output torque Tout will
decrease. This is represented by the section of A-to-B in Fig. 3.52.
The torque of C1 will drop to zero as the capacity of C2 reaches its critical capacity. The critical
capacity of C2 is defined by Eq. 3.55, obtained by setting Tn1 to zero in Eq. 3.52. At this point,
the clutch C2 is ready to take the entire load of the input torque, and the capacity of C1 can
be, and has to be, reduced to zero. The output torque, at same time, will drop to the level equal
to the second gear output torque level. This can be proven by substituting the critical capacity
equation (Eq. 3.55) into Eq. 3.54. This bottom point of output torque is the B point shown
in Fig. 3.52.
106
6293_Book.indb 106 1/11/13 3:53 PM

Fig. 3.51 Lever diagram of a two-speed gear train.
Fig. 3.52 Power-on up-shift from first gear to second gear.
In the entire torque phase up to this point, the capacity of C1 has to be greater or equal to its
load torque described by Eq. 3.52, so C1 will not slip. In addition, because node N1 will start to
pick up positive speed from this point, the capacity of C1 has to become zero from this point
onward. Otherwise, a negative torque will be developed on C1 that will further decrease the
output torque. The key here is that as the oncoming clutch C2 reaches its critical capacity, the
capacity of the off-going clutch C1 has to become zero—not early, not late. If early, the input
speed will flare (engine speed increasing in an uncontrolled fashion). If late, the clutches will
tie-up. In both cases, the output torque will drop further lower from the theoretical ideal
B point.
The torque phase is ended when the oncoming clutch C2 reaches its critical capacity, and con-
sequently the off-going clutch C1 is completely released. At this point the net torque exerted
on the input shaft is zero. Therefore, no input speed change could occur. To bring the input
speed to the second gear level, the capacity of C2 will be continually increased to the level at
which the input speed decelerates at a desired rate. This is the beginning of the inertial phase.
In the inertial phase, the speed change occurs. The input speed decreases from first gear speed
level to second gear speed level, the slip speed of clutch C2 increases from negative to zero, and
the slip speed of clutch C1 increases from zero to a positive value. To accomplish this speed
change, an extra torque has to be applied by clutch C2. This torque is called the inertial torque,
which can be calculated from the desired input speed deceleration rate and the inertia of the
input shaft. The C2 clutch torque and the output torque during the inertial phase are given by
107
6293_Book.indb 107 1/11/13 3:53 PM

Chapter 3
Eqs. 3.56 and 3.57. In Eqs. 3.56 and 3.57, a is the desired deceleration of the input shaft, and
Iin is the inertia of the input shaft. Equations 3.56 and 3.57 are obtained by applying the prin-
ciple of balance of moment about the output node and node N1, respectively, with the inertial
torque Iina added to the input node. In the inertial phase, the output torque will increase as
shown in Fig. 3.52; this is due to the inertial torque. During the inertial phase, the input speed
will continually decelerate until the slip speed of the oncoming clutch C2 becomes zero, and
therefore the input speed reaches the second gear level. Once the speed change is complete, the
inertial torque disappears; that is, a becomes zero. Therefore, from Eqs. 3.56 and 3.57, it can
be seen that the output torque and the C2 clutch torque drop to the second gear level. In the
inertial phase, a closed-loop control can be applied to maintain the desired input speed decel-
eration rate.
Note that the above description of power-on up-shift mechanics assumes that the input torque
is constant during gear shifting; there is no control over the input torque.
Tn1b = Tina − Tn 2 (b + c ) (3.52)
Tout = Tn1 + Tn 2 + Tin (3.53)
a +b c r −r
Tout = Tin − Tn 2 = Tinr1 − Tn 2 1 2
b b r2 − 1
(3.54)
Here, r1 and r2 are the first and second gear ratio, respectively.
a
Tn 2 = Tin = Tin (r2 − 1) (3.55)
b +c
a
Tn 2 = (Tin − I inα ) = (Tin − I inα )(r2 − 1) (3.56)
b +c
a +b +c
Tout = (Tin − I inα ) = (Tin − I inα ) r2 (3.57)
b +c
The input speed change during the inertial phase of a power-on up-shift can also be controlled
by input torque reduction, as shown in Fig. 3.53. Here, in the inertial phase the oncoming
clutch capacity can only be brought to the critical capacity level, and input speed deceleration
is accomplished by reducing the input torque. This control methodology provides less output
torque disturbance than the control methodology given in Fig. 3.52.
Power-on up-shift control technologies have gone through three generations of evolution. The
first generation is illustrated in Fig. 3.52. In this control concept, there is no control of trans-
mission input torque during gear shifts. The input speed decrease (inertial phase) is achieved
by increasing the oncoming clutch torque capacity. Therefore, the output torque has a check
mark shape. The second generation of power-on up-shift control technology is illustrated in
Fig. 3.53. In this control method, the input speed change is achieved by input torque reduc-
tion during the inertial phase instead of increasing the oncoming clutch torque capacity. As a
result of this control methodology, the output torque disturbance during gear shift is reduced.
The third generation of power-on up-shift control technology is called constant output torque
108
6293_Book.indb 108 1/11/13 3:53 PM

Fig. 3.53 Power-on up-shift from first gear to second gear, with input torque reduction control.
power-on up-shift (also called arbitrary output torque shaping of power-on up-shift). The
details of this new control concept will be described in Section 3.7, which discusses integrated
powertrain control.
Now let’s consider power-on down-shift. Referring to Fig. 3.54, the power-on down-shift from
second gear to first gear is started by reducing the torque capacity of the off-going clutch C2 to
a level such that the off-going clutch will slip, and the input speed will accelerate at a desired
rate by the input torque. This is the inertial phase of the power-on down-shift (note that the
inertial phase comes first in a power-on down-shift). A closed-loop control of the C2 torque
capacity is applied to maintain the desired input speed acceleration during the inertial phase.
As the slip speed of the oncoming clutch C1 reaches zero, the torque capacity of the oncoming
clutch C1 will be rapidly increased to it fully engaged level, and the torque capacity of the
off-going clutch C2 will be rapidly reduced to zero. This second phase is the torque phase
of power-on down-shifts. Note: having a well-designed hydraulic clutch control system that
Fig. 3.54 Power-on down-shift from second gear to first gear.
109
6293_Book.indb 109 1/11/13 3:53 PM

Chapter 3
enables the rapid increase or decrease of clutch capacity with the right timing in a controlled
manner is very important for achieving a smooth power-on down-shift.
The gear-shifting mechanics discussed in this section can be simulated by using the generic
dual-clutch model presented in the earlier section.
3.2.2 Hydraulic control system for clutch-to-clutch

shift controls
The torque capacity of gear-shifting clutches in clutch-to-clutch shift automatic transmis-
sions is controlled by electrohydraulic pressure control systems, which were described in the
previous section. Hydraulic control systems for clutch-to-clutch shift automatic transmissions
generally have a configuration as shown in Fig. 3.55 [3-2, 3-3, 3-4]. In this configuration, there
is a group of electrohydraulic pressure control valves and a group of hydraulic pistons for
clutch actuation. Between these two groups, a logic valve arrangement connects electrohydrau-
lic pressure control valves to the corresponding hydraulic pistons, while it provides the desired
failure mode responses. Failure mode responses usually include prevention of unintended
change of vehicle moving directions, transition into safer states in case of failure of electrical
power, and some level of limp-home capability.
3.2.3 Dynamic simulation model for studying

clutch-to-clutch shift controls
A simulation model for studying clutch-to-clutch shift controls can be constructed by combin-
ing the generic dual-clutch model with the hydraulic clutch control system model introduced
in previous sections.
3.2.4 Control of power-on up-shifts

Based on the gear-shift mechanics described in the previous section, controls of power-on up-
shifts are divided into four phases: fill phase, torque phase, inertial phase, and end phase.
Fig. 3.55 Hydraulic control system for clutch-to-clutch shift transmissions.
110
6293_Book.indb 110 1/11/13 3:53 PM

First, let’s consider control of clutch fill phase. During the fill phase, the piston of the hydrau-
lic cylinder of the oncoming clutch is stroked forward to eliminate the gap between the piston
and clutch plates. At the end of the fill phase, the clutch plates are pressed to each other, but
with near-zero torque capacity. To do so, a clutch fill command is applied to the oncoming
clutch pressure control solenoid. The “signal of clutch fill” command, as shown in Fig. 3.56,
has a fill pulse with a pressure level of Pfill and time duration of Tfill, followed by a low-pres-
sure-level command of Phold with time duration of Thold. Phold provides a force just above the
clutch-return-spring force, and provides a near-zero clutch torque capacity. The key parameter
here is the fill time Tfill. Tfill usually is temperature compensated, and is adapted using the fill-
time learning algorithm. Several fill-time learning algorithms have been developed, including
turbine-speed pull-down time based, output speed deceleration based, pressure sensor based,
and pressure switch based.
Second, let’s consider the torque phase control and off-going clutch control methods. The
torque phase follows the fill phase. In a torque phase, the oncoming clutch pressure is ramped
up to the level of Pinertia with a certain rate. The off-going clutch pressure is dropped to zero as
the oncoming clutch reaches its critical capacity. The key here is to reduce the off-going clutch
pressure to zero at the correct timing. Too early will result in engine speed flare, and too late
will result in transmission tie-up.
Several off-going clutch release control methodologies have been developed. The first type
is the model-based open-loop control method. In this method, the off-going clutch pressure
command is scheduled in alignment with the oncoming clutch pressure ramp-up command,
such that before the oncoming clutch reaches its critical capacity, the off-going clutch does
not slip, and as the oncoming clutch reaches its critical capacity, the off-going clutch capacity
reaches zero. The method uses the steady-state powertrain model to determine the required
off-going clutch capacity during the oncoming clutch pressure ramp-up, and to determine
when the oncoming clutch will reach its critical capacity.
The second method of off-going clutch release control is the output speed acceleration-based
method. As described in the gear-shifting mechanics section, during the torque phase, as the
oncoming clutch capacity increases, the output torque will decrease from point A to point
B, as shown in Fig. 3.52. Therefore, output speed acceleration will decrease from one level to
another level corresponding to the torque points A and B, respectively. This change of output
speed acceleration can be used to determine the progress of the torque phase, and therefore
can be used to determine the control signal of the off-going clutch.
The third method of off-going clutch release control is the closed-loop off-going clutch slip
control method. In this control method, off-going clutch pressure is reduced and then con-
trolled by a closed loop; the objective of this closed loop control is to maintain the slip speed
of the off-going clutch at a desired value (usually around 20 rpm). As the oncoming clutch
capacity increases, the required off-going clutch capacity for maintaining the desired slip speed
will decrease. Once the oncoming clutch reaches its critical capacity, the closed-loop capacity
control of the off-going clutch automatically reduces the off-going clutch capacity to zero.
Now let’s consider the inertial phase and the speed-change control method. The end of
the torque phase and the beginning of the inertial phase are marked by the oncoming clutch
reaching its critical capacity, and the off-going clutch reaching zero capacity. In the inertial
phase of power-on up-shifts, input speed changes such that the slip speed of the oncoming
clutch becomes zero. There are three control methods to accomplish input speed change. The
first method uses the oncoming clutch as a control element. The oncoming clutch capacity
is increased from its critical capacity to provide clutch torque to reduce the engine speed. In
111
6293_Book.indb 111 1/11/13 3:53 PM

Chapter 3
this method there is no active control of engine torque. A closed-loop control is used to let
the input speed change follow a desired profile, and therefore, to provide a soft clutch lock.
The second method is the engine torque control method. In this method, the oncoming clutch
capacity is maintained at its critical capacity. The engine speed change and soft clutch lock
are controlled by controlling engine torque reduction. The third method is a combination of
oncoming clutch control and engine torque reduction control.
Once oncoming clutch slip becomes zero, control enters the end phase. In the end phase, the
oncoming clutch pressure is increased to ensure firm engagement.
3.2.5 Control of power-on down-shifts

Based on the descriptions of the power-on down-shift mechanics, a power-on down-shift
control starts from reduction of the off-going clutch pressure to the level that provides a
desired engine speed increase rate. At the same time, the oncoming clutch is filled and is held
at near-zero-capacity position. When the oncoming clutch slip nears zero, the oncoming clutch
pressure is quickly increased to the fully engaged level, and the off-going clutch pressure is
quickly reduced to zero. Closed-loop control could be applied to control the off-going clutch
pressure to achieve the desired engine speed increase rate.
3.2.6 Hydraulic clutch control system design

The oncoming and off-going clutches are controlled by hydraulic clutch control systems, which
consist of a hydraulic piston, pressure regulation valves, and PPC solenoids. The dynamic char-
acteristics of clutch control systems have determinative effects on the quality of gear-shifting
control. A good clutch control system design is characterized by the following features:
• Be able to fill the clutch quickly and consistently without pressure overshot.
• Be able to hold the clutch at near-zero clutch capacity consistently.
• Clutch capacity follows command responsively after the fill phase with minimal tracking
error, steady-state error, and hysteresis.
Fig. 3.56 Oncoming and off-going clutch pressure control of power-on up-shifts.
112
6293_Book.indb 112 1/11/13 3:53 PM

The optimal design of hydraulic clutch control systems has been discussed in Section 3.1.5.
This section summarizes some configurations of hydraulic clutch control systems, with a brief
discussion of their characteristics.
The first configuration is shown in Fig. 3.57. This is the most popular configuration. There
is a clutch feed orifice between the pressure regulation valve and the hydraulic piston. This
orifice is the dominant hydraulic resistance, so that the clutch fill time can be controlled more
consistently. The clutch compliance is minimized, and therefore the clutch-apply force is more
responsive to the command. In this configuration, the two-stage pressure control system
(PPC solenoid plus pressure regulation valve) could be replaced by a high-flow, direct-acting
pressure control valve, as described in an earlier section.
The second configuration is shown in Fig. 3.58. In this configuration, an orifice is placed at the
line pressure feed to the pressure regulation valve. Therefore, the pressure regulation valve acts
as a trim valve during the clutch fill phase. The existence of this line pressure feed orifice can
cause pressure overshot at the end of the clutch fill phase. To eliminate the pressure overshoot,
a wave plate is placed between the piston and the clutch plate. However, the presence of the
wave plate decreases the responsiveness of the system. Furthermore, when the wave plate is not
fully compressed, the clutch-apply force is not determined by the hydraulic pressure directly,
but by the displacement of the piston. Due to the friction force between the hydraulic piston
and cylinder, a hysteresis loop in the piston position is created in response to the hydraulic
pressure. Therefore, accurate and consistent control of the clutch-apply force is difficult. Wave
plates also increase the axial dimension of transmissions.
Fig. 3.57 Clutch control system with a clutch feed orifice and minimized clutch compliance.
Fig. 3.58 Clutch control system with a line feed orifice and a wave plate to increase clutch compliance.
113
6293_Book.indb 113 1/11/13 3:53 PM

Chapter 3
Fig. 3.59 Clutch control system using accumulator to increase clutch compliance.
An alternative method to increase clutch compliance is to use an accumulator, as shown in

Fig. 3.59. This method eliminates the negative effects of wave plates while providing
additional compliance.
3.2.7 Clutch fill detection

As described previously, filling the clutch (i.e., stroking the clutch piston) speedily, accurately,
and consistently is one of the most important control tasks for achieving smooth and quick
clutch-to-clutch shifts. The major control parameter of clutch fill control is the fill time. The
fill time command Tfill is an adaptive control parameter. After each shift, the adaptive algo-
rithms evaluate shift results and determine when the clutch is actually filled, and then adjust
the adaptive control parameters accordingly to improve the future clutch fill control. In the
fill time adaptive algorithms, detection of the actual clutch fill, of course, is the most import
technology.
Sensors such as pressure transducers and pressure switches can be used to detect clutch fill;
however, they increase the cost of the product. A technology using the existing transmis-
sion output speed signal to detect clutch fill has been developed. As described in the previous
section, for a power-on up-shift, as the oncoming clutch is filled and starts to gain torque
capacity, the transmission output torque will start to decrease; therefore, the angular accelera-
tion of the transmission output shaft will start to decrease as well. By detecting the change of
the angular acceleration of the transmission output shaft, the end of clutch fill and the begin-
ning of the torque phase can be determined. The angular acceleration can be obtained from the
transmission output speed signal using acceleration estimators described in the next section.
3.2.8 Acceleration estimator

Algorithms for calculating acceleration from measured speed signals are called acceleration
estimators. Due to the existence of system and measurement noise, directly taking derivatives
of a speed signal will produce a very poor acceleration signal. Therefore, anti-noise capability
is a must for a practical acceleration estimator. Two such acceleration estimators are intro-
duced next.
3.2.8.1 Kalman acceleration estimator

The first type includes acceleration estimator designs based on Kalman estimator theory [3-5].
The discrete plant equations for designing the Kalman acceleration estimator are given in
Eq. 3.58.
114
6293_Book.indb 114 1/11/13 3:53 PM

x (n + 1) = Ax (n) + Gw (n)
(3.58)
y (n) = Cx (n) + v (n)
Here, x is the system state vector defined as one shown in Eq. 3.59, with w as angular velocity,
a as angular acceleration, and J as angular jerk.
 ω 
  (3.59)
x = α 
 J 
A, C, and G are the state matrix, measurement matrix, and state noise weight matrix, respec-
tively; they are defined as follows in Eqs. 3.60–3.62. Here, T is the sampling period.
 1 T 0 
 
A = 0 1 T  (3.60)
 0 0 1 
C =  1 0 0  (3.61)
 g 
 1 
G =  g2  (3.62)
 
 g 3 
Also, w and v are the white process and measurement noise; they satisfy:
E(w) = W(v) = 0, E(ww) = q, E(vv) = r, E(wv) = 0
Here, g1, g2, and g3 are the weights of process noise w for each state variable, respectively. Then,
the equation of the Kalman acceleration estimator, which is designed based on the previous
plant model, is shown in Eq. 3.63.
xˆ(n + 1) = Axˆ(n) + L ( y (n) − Cxˆ (n)) (3.63)
Here, x̂ is the estimated state vector. L is the Kalman estimator gain vector; it is obtained from
Eq. 3.64.
L = PCT r −1 (3.64)
Here, P is obtained from solving the algebraic Riccati equation, Eq. 3.65.
AP + PAT − PCT r −1 CP + GqGT = 0 (3.65)
MATLAB is a tool for calculating the Kalman estimator gain L. The MATLAB command for
doing so is as follows.
115
6293_Book.indb 115 1/11/13 3:53 PM

Chapter 3
[Kest, L, P] = kalman(ss(A, G, C, 0, T), q, r, 0)
This command returns the Kalman estimator gain vector L, the covariance P, and the Kalman
estimator system equations, Kest.
The block diagram of the Kalman acceleration estimator is shown in Fig. 3.60. Here, L1, L2,
and L3 are the first, second, and third elements of the Kalman estimator gain vector L. T is the
sampling period, omega is the estimated angular velocity signal, alpha is the estimated angular
acceleration signal, and y is the measured angular velocity signal.
For the reader’s convenience, Kalman estimator gains designed for various sampling periods T
are given in Table 3.1.
Figure 3.61 is the comparison of acceleration signals obtained from the Kalman acceleration
estimator and obtained from the direct derivative calculation. In Fig. 3.61, the first graph is the
measured angular velocity signal, the second graph is the angular acceleration obtained from
the direct derivative calculation, and the third graph is the angular acceleration obtained from
the Kalman acceleration estimator. It can be seen that the Kalman estimator provides a much
better result.
3.2.8.2 Filter-observer based acceleration estimator

This section introduces another acceleration estimator. This acceleration estimator is con-
structed by merging a low-pass filter with a state observer, and therefore is named the FO
(Filter Observer) acceleration estimator. The block diagram of the FO acceleration estimator is
shown in Fig. 3.62. In the figure, y is the measured angular velocity signal, omega is the filtered
Fig. 3.60 Block diagram of Kalman acceleration estimator.
Table 3.1 Kalman Acceleration Estimator Gains

T L1 L2 L3
0.003125 s 1.6160 7.1517 0.1430
0.00625 s 1.1810 7.1720 0.1434
0.025 s 1.3034 7.2979 0.1459
116
6293_Book.indb 116 1/11/13 3:53 PM

Fig. 3.61 Comparison of the Kalman acceleration estimator and direct derivative calculation.
Fig. 3.62 Block diagram of the filter-observer acceleration estimator.
angular velocity signal, and alpha is the estimated angular acceleration signal. The first portion
of the FO acceleration estimator is a low-pass filter. The gain Kf determines the bandwidth of
this low-pass filter. The output of the low-pass filter, omega (filtered angular velocity), is fed to
the second portion, the state observer. The output of the state observer is alpha, the estimated
angular acceleration. Two trapezoidal discrete-time integrators are used in the state observer.
The proportional gain Kp and integral gain Ki determine the converging speed and stability
of the state observer. A simplified version of the FO acceleration estimator is shown in Fig. 3.63.
In this version, only proportional gain Kp is used to achieve the desired converging speed and
stability.
Figure 3.64 compares the FO estimator and the Kalman estimator. It can be seen that the FO
acceleration estimator speeds up the estimation without increasing the noise level.
Another thing worth pointing out is that calibration of the FO estimator is more transparent
and straightforward. Kf determines the bandwidth of the low-pass filter. The PI gains, Kp and
Ki, determine the stability and converging speed of the state observer.
117
6293_Book.indb 117 1/11/13 3:53 PM

Chapter 3
Fig. 3.63 Comparison of the Kalman acceleration estimator and direct derivative calculation.
Fig. 3.64 Comparisons of the FO estimator and the Kalman estimator.
3.2.9 Canceled shifts, transitional shifts, and

double-transition shifts
In real-world driving, the driver could step-into or back-out from the accelerator pedal at
any time, even when a gear shift is in progress. In such cases, canceling the current gear shift
(canceled shift), or transition to a different type of gear shift to achieve the new demanded
gear (transitional shift) could occur. Also, in the case of the accelerator pedal position having a
large step change, a multiple-step skipped gear shift could be commanded. Such a shift usually
requires two oncoming clutches and two off-going clutches (double-transition shift). This
section provides a brief discussion on how to handle these gear shifts. The intention of the dis-
cussion is to provide “a way of thinking’” for designing control strategies.
First let’s consider the cancelation of a power-on up-shift. Cancelation of power-on up-shifts
happens when the driver further steps-into the accelerator pedal during a power-on up-shift.
Therefore, the cancelation of a power-on up-shift is also a power-on shift control event. The
control actions required to cancel a power-on up-shift depend on the phase the current shift is
in. Table 3.2 shows the control actions to cancel a power-on up-shift when in different phases.
The first row shows which phase the control is in when the cancel shift is commanded. The
second row shows the action to take to cancel the shift.
118
6293_Book.indb 118 1/11/13 3:53 PM

Table 3.2 Control Actions to Cancel a Power-on Up-shift

Current Power-on
Up-shift Phase Fill Phase Torque Phase Inertial Phase
Action to Take to Keep the off-going Increase the off- Switch to the
Cancel the Power-on clutch engaged. going clutch inertial phase of
Up-shift capacity and power-on down-
Stop fill of oncoming
re-engage it. shift control. The
clutch, and release
original oncoming
it. Reduce the
clutch becomes
oncoming clutch
off-going clutch,
capacity and release
and the original
it.
off-going clutch
becomes oncoming
clutch. Complete
the shift based
on the power-on
down-shift control
methodology.
Cancelation of power-on down-shifts usually occurs when the driver releases the accelerator
pedal during the gear shift. Table 3.3 shows the actions required to cancel a power-on down-
shift when in different phases.
Now let’s consider transitional shifts. Transitional shifts occur when the desired gear becomes
different from the currently targeted gear. When the desired gear changes during a gear shift,
the required actions to transition to the new gear shift control to achieve the new targeted
gear depend on the clutching chart of the transmission. Let’s use an eight-speed automatic
transmission as an example to consider the control strategy of transitional shift controls. Table
3.4 is the clutching chart of the example transmission, and Fig. 3.65 is the stick diagram of the
transmission.
Let’s consider the transitional shift from 6-5 to 6-4 or from 6-5 to 6-3. From the clutching
chart, it can be seen that these down-shifts have the same oncoming clutch (C12345R), but
different off-going clutches. For the 6-5 down-shift, the off-going clutch is C23468; for the 6-4
Table 3.3 Control Actions to Cancel a Power-on Down-shift

Current Power-on Down-
shift Phase Inertial Phase Torque Phase
Action to Take to Cancel the Release the oncoming clutch The torque phase of a
Power-on Down-shift from its fill and hold phase. power-on down-shift is
usually a short and fast
Smoothly re-engage the off-
control phase, and is
going clutch.
difficult to cancel. Therefore,
complete the power-on
down-shift, and initiate the
next up-shift.
119
6293_Book.indb 119 1/11/13 3:53 PM

Chapter 3
Table 3.4 Clutch Chart of an Eight-speed Transmission

CB1278R CB12345R C23468 C13567 C45678R
1st X X X
2nd X X X
3rd X X X
4th X X X
5th X X X
6th X X X
7th X X X
8th X X X
Rev X X X
down-shift, the off-going clutch is C13567; and for the 6-3 down-shift, the off-going clutch
is C45678R. So the transitional shift from 6-5 to 6-4 or from 6-5 to 6-3 requires transition
control of off-going clutches. The primary objective of such transition control is to maintain
the same input speed acceleration rate without interruption, while switching the off-going
clutches.
The transitional shift from 2-3 to 2-4 requires change of oncoming clutch, while the transi-
tional shift from 2-3 to 2-5 could be handled easier by completing the 2-3 shift first, and then
initiating the 3-5 shift.
While some of the other transitional shifts of the example eight-speed automatic transmissions
are similar to those just described, some are different. The way to control these transitional
shifts could be determined by considering the change in the oncoming and off-going clutches
based on the clutching chart.
Finally, let’s consider double-transition shifts. As the result of the large step change of
the accelerator pedal position, a multi-step skip shift could be commanded. For example,
on step-in of the accelerator pedal from zero to the full depressed position, a 7-4 or a 7-3
Fig. 3.65 Stick diagram of an eight-speed transmission.
120
6293_Book.indb 120 1/11/13 3:54 PM

down-shift could be commanded. Both 7-4 and 7-3 require two off-going and two oncoming
clutches. Such shifts are called double-transition shifts.
The double-transition shifts are controlled by breaking it into two single-transition shifts. The
7-4 shift could be broken into 7-5 and 5-4 shifts, and the 7-3 could be broken into 7-5 and
5-3 shifts. The key of controlling such double-transition shifts is to have a seamless continuity
between the two single-transition shifts.
3.3 Electronic torque converter clutch control

Torque converters are excellent devices to launch a vehicle or to gain torque multiplication.
However, when a torque converter reaches its coupling point, although the torque multiplica-
tion is diminished, a slip speed of around 300 rpm still exists between its pump and turbine
shaft. This large slip speed and torque between pump and turbine shaft result in a consider-
able amount of power loss. To reduce this power loss, a frictional clutch is placed in the torque
converter. When this clutch is engaged, the pump and turbine shafts will be locked together
(the clutch short circuits the fluid path), and the power loss caused by the torque converter slip
is eliminated.
Engine torque is not smooth due to the cylinder combustion and reciprocal motion of the
pistons. If the torque converter clutch (TCC) is fully locked, the engine torque pulsation will
be transmitted to the driveline and vehicle body. To avoid this, the TCC usually is not fully
locked. The torque capacity of the TCC is controlled to lower the converter slip speed. The
lower slip speed (usually between 5 and 20 rpm) significantly reduces the power lost in com-
parison with the 300 rpm converter slip speed without TCC control. This section is devoted to
the torque converter clutch controls. First, the hydraulic control system of the torque converter
and torque converter clutch is presented, and then the electronic control algorithm for TCC
slip speed control will be discussed.
3.3.1 Hydraulic system for torque converter

clutch control
A hydraulic control system for torque converter clutch control is shown in Fig. 3.66. The TCC
system depicted is an elegant two-path system that multiplexes the friction interface as both
torque-carrying member and piston. Under deenergized conditions, the TTC Release pressure
is higher than the apply pressure. This pressure imbalance results in flow across the TCC
interface and prevents the application of the clutch. When this balance is changed to favor an
apply pressure that is higher than the release pressure, the TCC is “applied,” resulting in a con-
trolled slip.
In Fig. 3.66, when the system is in the TCC release state, the TCC control valve moves to the
left side, to feed the TCC feed limit oil to the TCC release port, and directs the oil coming
back from the TCC apply port to the cooler. When the system is in TCC apply state, the
TCC control valve moves to the right side, to feed the Reg apply oil to the TCC apply port,
and directs the TCC feed limit oil to the cooler. The TCC Reg apply valve controls the TCC
apply pressure. The PWM solenoid controls the system-state (release state or apply state)
and the TCC apply pressure. An electronic controller controls the PWM solenoids based
upon the vehicle operation. Although the pressure control solenoid shown here is a PWM
solenoid, more recently, proportional pressure control solenoids are being used in this type
of application.
121
6293_Book.indb 121 1/11/13 3:54 PM

Chapter 3
Fig. 3.66 Torque converter clutch control system.
3.3.2 Electronic control algorithm for torque converter

clutch control
There are four control modes in a TCC control algorithm. They are released mode, applying
mode, releasing mode, and ECCC (Electronic Converter Clutch Control) mode. The mode
transition diagram of TCC control is shown in Fig. 3.67. In the released mode, TCC is released
by commanding 0% duty cycle to the PWM solenoid; therefore, no torque is transmitted
through the torque converter clutch. When the TCC apply condition is satisfied, control tran-
sitions to the applying mode. In the applying mode, TCC is brought from the released state
to the applying state smoothly by gradually increasing the duty cycle command to the PWM
solenoid. As a result of applying TCC, the converter slip decreases. As converter slip speed
becomes lower than a predetermined value, control transitions to the ECCC mode. In ECCC
mode, the TCC apply pressure is controlled by a feedback and feed-forward control algorithm
to maintain converter slip at a predetermined speed (usually 5–20 rpm, and is determined
based on engine characteristics, dynamic characteristic of torque converter damper, and the
TCC hydraulic control system). If the TCC release condition is satisfied while in the ECCC
mode, control transitions to the releasing mode. In the releasing mode, the TCC is brought
from the ECCC state to the released state smoothly by gradually decreasing PWM duty cycle
command to zero, and then entering the released mode. The TCC apply and release condition
usually is based on vehicle speed. There is a TCC apply vehicle speed and a TCC release vehicle
speed. A hysteresis between these two speeds is necessary (apply speed > release speed) to
prevent apply-release hunting. TCC apply condition is “vehicle speed > apply speed,” and TCC
release condition is “vehicle speed < release speed.” TCC apply and release speeds usually are
functions of engine power level.
In the ECCC mode, a feedback/feed-forward control is used to control the TCC slip speed. The
block diagram for the ECCC mode control is shown in Fig. 3.68. In the block diagram, Dwr
122
6293_Book.indb 122 1/11/13 3:54 PM

Fig. 3.67 Mode transition diagram of TCC control.
Fig. 3.68 Block diagram of an ECCC control system.
is the desired slip speed, Dw is the actual slip speed, Te_cal is the estimated engine torque that
usually is computed by the engine controller, Te is the actual engine torque, C1 is the feedback
controller, and C2 is the feed-forward controller. The feedback controller C1 can be a PI
controller or other type of feedback controller that provides zero steady-state error. The feed-
forward controller C2 usually is a gain control type.
A Simulink model of the TCC control algorithm is shown in Figs. 3.69–3.73. Figure 3.69 is the
main model. The other four figures are the submodels of the four control modes. The mode
transition is implemented in the main model using the switch case and case action mecha-
nism. The input to the switch case block is the variable, mode. Based on a mode value of 1,
2, 3, or 4, the submodel of released mode, applying mode, ECCC mode, or releasing mode is
activated, respectively. The control action and mode transition logic are implemented in each
submodel. For example, in the submodel of ECCC Mode shown in Fig. 3.73, the upper portion
is the mode-transition logic. When the vehicle speed variable (veh_spd) becomes lower than
the TCC release speed (TCC_release_spd), the mode variable (mode) is set to 4 to switch to
releasing mode; otherwise, the mode variable (mode) is set to 3 to stay in ECCC mode. The
lower portion of the submodel is the control action model that implements the feed-forward
and feedback control of TCC slip speed.
123
6293_Book.indb 123 1/11/13 3:54 PM

Chapter 3
Fig. 3.69 Simulink model of the TCC control algorithm (main model).
Fig. 3.70 Submodel of released mode.
124
6293_Book.indb 124 1/11/13 3:54 PM

Fig. 3.71 Submodel of applying mode.
Fig. 3.72 Submodel of releasing mode.
125
6293_Book.indb 125 1/11/13 3:54 PM

Chapter 3
Fig. 3.73 Submodel of ECCC mode.
3.4 Dynamic analysis of torque converter

clutch damper
For some applications, it is desirable to lock up the torque converter clutch, as opposed to
slipping the clutch, to provide a manual-transmission-like feeling or further improve fuel
economy. When the TCC is locked, the torque converter clutch damper must provide the
desired level of torsional isolation to prevent the engine torque pulsation from being transmit-
ted to the vehicle body. Also, a better damper will allow more-aggressive operation of TCC
such as lower slip speed or early TCC application.
The torque converter clutch damper is a mass and spring torsional isolation device. To analyze
torque converter clutch damper designs, a dynamic simulation model of the system, as shown
in Fig. 3.74, is necessary. In this system, the engine is modeled as a torque source with har-
monics. The engine harmonic torque model is given in Eq. 3.66.
n
Te = A0 + ∑ Ai sin (αiθe + ψi ) (3.66)
i =1
126
6293_Book.indb 126 1/11/13 3:54 PM

Fig. 3.74 System for TCC damper study.
Here, ai is the harmonic number; Ai is the magnitude for the harmonic number ai, yi is the
phase shift for the harmonic number ai, qe is the angular position of the engine shaft, and Te is
the engine torque. For six-cylinder engines, ai takes the number of 3, 6, 9, 12, and so on. For
eight-cylinder engines, ai takes the number of 4, 8, 12, 16, etc.
The inertia Ii in Fig. 3.74 includes engine, flex plate, and pump inertias. There are three torques
acting on Ii: engine torque Te, pump torque Tp, and TCC torque Ttcc. The angular velocity
of Ii is we. TCC can be modeled as a frictional clutch using the hyperbolic tangent function
described in an earlier section. The torque converter can be model using torque ratio and
k-factor, also discussed earlier. The TCC damper is a spring and friction system, and can be
modeled by Eq. 3.67. Here, kd is the spring constant of the TCC damper, and Tf is the friction
force of the TCC damper.
Td = kd ∫ (ω d − ωt )dt + T f tanh ((ωd − ωt ) / α ) (3.67)
The inertia It includes turbine and transmission inertias. The gear ratio rg and final drive ratio
rf provide the relationships, as shown in Eqs. 3.68–3.71.
Ti = Ta / rg (3.68)
ωo = ωi /rg (3.69)
Tw = Ta ⋅ rf (3.70)
ωa = ωw ⋅ r f (3.71)
127
6293_Book.indb 127 1/11/13 3:54 PM

Chapter 3
The axle and tire compliance can be modeled by Eq. 3.72. Here, ka is the spring constant, and
Da is the damping coefficient.
Ta = ka ∫ (ω o − ωa )dt + Da (ωd − ωt ) (3.72)
The inertia Iv includes the inertia of the vehicle wheel and the vehicle mass. Id is the inertia of
the damper plate.
The equation of motion of inertias Ii, Id, It, and Iv are shown in Eqs. 3.73–3.76.
ω e = (Te − Ttcc − Tp ) / I i (3.73)
ω d = (Ttcc − Td ) / I d (3.74)
ω t = (Tt + Td − Ti ) / I t (3.75)
ω w = (Tw − Dw ωw ) / I v (3.76)
With the previous information, a simulation model can be developed, and analysis of different
TCC damper designs can be performed. Figure 3.75 shows the simulation results of two differ-
ent TCC damper designs. The one on the left is the standard damper, and the one on the right
is a damper with a lower rate and longer travel spring (softer damper). It can be seen that the
later one significantly reduced the fluctuation of the torque converter output torque in com-
parison with the standard damper. However, a softer damper demands a more-complicated
mechanical design and more space.
Fig. 3.75 Simulation comparison of two different damper designs.
128
6293_Book.indb 128 1/11/13 3:54 PM

3.5 Friction launch control

Torque converters are widely used as a vehicle launch device traditionally. Although torque
converters provide excellent vehicle launch performance, they are expensive and take up a
large amount of space. Recently more and more transmission designs use a friction clutch as
a launch device to reduce cost and packaging space. An example of a friction launch system is
illustrated in Fig. 3.76. The system uses gear-shifting clutches as the launch devices. The clutch
C1234 is the forward launch clutch, and the clutch C23R is the reverse launch clutch. Other
examples of transmission systems that make use of a friction clutch as a vehicle launch device
include the Dual-Clutch Transmission (DCT) and Automated Manual Transmission (AMT).
This section discusses controls of friction launch.
Typical requirements for friction launch control systems are:
• Idle: when throttle is closed, brake is applied, and vehicle is stopped, the friction launch
clutch should have zero capacity.
• Creep: when the brake is released and throttle is closed, the friction launch clutch should
provide a certain level of creep torque.
• Launch: when the driver depresses the accelerator, the control should provide smooth
vehicle launch at a desired engine stall speed. The engine stall speed should be a function
of throttle opening.
• Smooth operation at any engine power level.
• Robust against system variations such as temperature and characteristics of frictional
materials.
• Smooth transition for tip in/out operation.
• Be able to control launch when vehicle is rolling in opposite direction without using bi-
directional speed sensing technologies.
Fig. 3.76 Diagram of a vehicle launch control system.
129
6293_Book.indb 129 1/11/13 3:54 PM

Chapter 3
Several different friction launch control algorithms have been developed. Typically, they can be
categorized into the following four types of control concepts.
• Type-1 is the closed-loop clutch-slip-speed-profile control concept.

• Type-2 is the closed-loop engine-speed-profile control concept.
• Type-3 is the torque converter emulation control concept.
• Type-4 is the engine loader control concept.
In the following paragraphs, the type-4 control concept is described. The type-4 concept, the
engine loader control concept, focuses on applying the correct load to the engine by control-
ling the launch clutch capacity, instead of focusing on vehicle movement.
The engine loader control concept commands the launch clutch torque as a function of engine
speed in a prescribed way, as illustrated by the launch clutch torque curve shown in Fig. 3.77.
The launch clutch torque curve intercepts the engine torque curve. The intercepting point
determines the engine stall speed.
Separate launch clutch torque curves are assigned to each throttle opening, as shown in
Fig. 3.78. In Fig. 3.78, Te is the engine torque curve, and Tc is the launch clutch torque curve.
The intercept points between engine torque curves and launch clutch torque curves of the
same throttle opening determine the stall speed for that throttle opening level. Therefore,
the optimal stall speed can be calibrated for each throttle opening to achieve a good tradeoff
Fig. 3.77 Launch

clutch torque curve and
engine torque curve.
Fig. 3.78 Multiple launch

clutch torque curves and
engine torque curves.
130
6293_Book.indb 130 1/11/13 3:54 PM

between performance and fuel economy. The intercept point between the idle engine torque
curve, Tc(idle), and idle launch clutch torque curve, Te(idle), determines the creep torque.
The launch clutch torque command generated from the calculation is then multiplied by a
modifier. The modifier is a variable that is a function of throttle opening and speed ratio. Speed
ratio is defined as the ratio of equivalent turbine speed to engine speed. The equivalent turbine
speed equals the product of the transmission output speed and transmission current gear ratio.
The value of the modifier ranges from 0 to 1. A typical modifier function is shown in Fig. 3.79.
The purpose of the modifier is to decrease the launch clutch torque as the clutch approaches
the lockup point, therefore resulting in a soft lockup.
The typical behavior of engine speed and equivalent turbine speed during the engine loader
control is shown in Fig. 3.80. The engine loader control has the following desired features:
• The engine stall speed can be freely and easily calibrated to the desired value for each
throttle opening.
• Because, in the early stage of the control, only the engine speed signal is used, the control
can handle launch naturally in the situation when the vehicle is rolling in the opposite
direction.
• For the same reason as above, the control is insensitive to output speed measurement
noise, which is especially high at low output speed.
• Engine loader control handles throttle tip-in and tip-out smoothly in a very simple way.
A Simulink model of the engine loader friction launch control algorithm is shown in Fig. 3.81.
In this model, the launch clutch torque Tcl is calculated in the function block “launch clutch
torque” with the equation shown in Eq. 3.77.
Fig. 3.79 Engine loader

modifier curves.
Fig. 3.80 Friction launch process

described in terms of engine speed
and equivalent turbine speed.
131
6293_Book.indb 131 1/11/13 3:54 PM

Chapter 3
Fig. 3.81 Block diagram of friction launch control algorithm.
Tcl = (Tloc k + Tlau (1 − k )) ⋅ g (3.77)
Here, g is the clutch torque gain, which is defined as the ratio of clutch torque to transmission
input torque. Tcl is the clutch torque command, which consists of two components: Tlau and
Tloc. Tlau is the launch torque, which is calculated based on the engine loader friction launch
control concept just described. Tloc is the clutch lock torque, which is used to lock up the clutch
after launch is complete. The transition from Tlau to Tloc is determined by the value of k, which
is the mode command that varies from 0 (launch) to 1 (lockup). When k equals 0, Tcl equals
Tlaug; when k equals 1, Tcl equals Tlocg. Because k changes from 0 to 1 following a prescribed
rate, transition from Tlau to Tloc is smooth. The calculation of Tlau, Tloc, and k are described in
the following paragraphs.
First, let’s describe the calculation of k. The value of k is the rate-limited value of the variable
“lockup” (see the block of lock rate limiter in Fig. 3.81). The variable lockup takes values of
0 (launch) or 1 (lock), and is determine by a state transition diagram “lock & unlock,” with
its details shown in Fig. 3.82. The initial state is the “Unlock” state. In the “Unlock” state,
the variable lockup is set to 0. When in the “Unlock” state, if the condition [output_speed >
lockup_speed] is satisfied, a state transition from “Unlock” to “Lock” is made, and the variable
lockup is set to 1. When in the “Lock” state, if the condition [output_speed < unlock_speed] is
satisfied, a state transition from “Lock” to “Unlock” is made, and the variable lockup is reset
132
6293_Book.indb 132 1/11/13 3:54 PM

Fig. 3.82 State transition diagram for determining launch clutch state.
to 0. The variable output_speed is the transmission output speed, and is often abbreviated as
TOSS. The variables lockup_speed and unlock_speed are functions of throttle opening. The lock
speed line and unlock speed line are shown in Fig. 3.83, together with 1-2 and 2-1 shift lines.
Second, let’s look at the calculation of Tlau. Tlau is the control variable that controls the friction
launch process. Tlau is a function of the throttle opening (TPS) and engine speed (TISS), and
is implemented by the 3-D look-up table “Tc vs. TPS & Tiss” in Fig. 3.81. A typical shape of
such functions is shown in Fig. 3.84. The line chart representation of this function is shown
in Fig. 3.85. Figure 3.85 also shows the engine torque curves as well as the launch clutch
torque curves. The launch torque curve for each throttle opening is designed to meet the
engine torque curve at a desired stall speed to obtain the optimal launch performance and
fuel economy. The launch torque calculated from the 3-D lookup table is then multiplied by
a modifier. The modifier is a function of throttle opening TPS and launch clutch speed ratio
SR, and is implemented by the 3-D look-up table “modifier vs. TPS & SR” in Fig. 3.81. SR is
defined as shown in Eq. 3.78.
TOSS ⋅ r
SR = (3.78)
TISS
Here, r is the transmission gear ratio during the launch. Figure 3.86 is the line chart of the 3-D
lookup table: modifier vs. TPS & SR. The throttle opening used by the 3-D lookup tables is
rate-limited and filtered, as shown in the block diagram of Fig. 3.81.
Fig. 3.83 Lockup and unlock speed line.
133
6293_Book.indb 133 1/11/13 3:54 PM

Chapter 3
With the control algorithm of Tlau, the friction launch process will be as shown in Fig. 3.87.
The engine speed first accelerates to the stall speed determined by the intercept point of the
launch clutch torque curve and engine torque curve. Then, the engine speed will stay at this
stall speed and wait for the “turbine” speed to catch up. When the clutch speed ratio nears the
coupling point, the modifier will kick in to reduce clutch lockup shocks.
Finally, let’s consider the calculation of Tloc. Tloc is calculated from engine torque, with a gain
and offset as shown at the top of Fig. 3.81.
Fig. 3.84 3-D lookup table to calculate friction launch torque as a function of throttle opening
and engine speed.
Fig. 3.85 Line chart representations of the 3-D lookup table to calculate friction launch torque as a
function of throttle opening and engine speed.
134
6293_Book.indb 134 1/11/13 3:54 PM

Fig. 3.86 Line chart for determining the value of modifier value as a function of speed ratio and throttle opening.
Fig. 3.87 Friction launch process achieved by the engine loader control strategy.
3.6 Shift scheduling system

The shift scheduling systems determine which gear a transmission should shift into based on
considerations of drive quality, drivability, and fuel economy. The standard shift scheduling
systems use shift-maps to determine the desired gear. The advanced shift scheduling systems
use artificial intelligence technologies to achieve more-adaptive, humanlike shift scheduling.
3.6.1 Shift map

An example of the throttle-position-based shift maps is shown in Fig. 3.88. The shift map
defines the up- and down-shift vehicle speeds as functions of throttle position. In the figure,
the lines labeled 1-2, 2-3, etc. are up-shift speed lines, and the lines labeled 2-1, 3-2, etc. are
down-shift speed lines.
135
6293_Book.indb 135 1/11/13 3:54 PM

Chapter 3
The up-shift lines affect the acceleration performance and fuel economy of a vehicle, and are
determined mainly from engine characteristics (engine torque and fuel consumption) and
drive quality requirements (response to driver’s demand). To understand the basic concept
of shift maps, let’s take a look at Fig. 3.89, which shows the vehicle traction efforts at 100%
and 50% throttle positions (TPS) for each gear ratio. To obtain maximum performance, up-
shifts shall occur at each intersection point. The up-shift speeds near 100% throttle opening
are chosen at these intersection points to achieve the best acceleration performance. In the
medium throttle position region, the shift speeds are chosen lower than these intersection
points, to gain some fuel economy, and to preserve some performance. In the low throttle
position region, the shift speeds are chosen to mainly maximize fuel economy.
The downshift lines are chosen as a result of tradeoffs between shift business and
torque reserve.
Fig. 3.88 Shift map for a six-speed transmission.
Fig. 3.89 Available traction effort curve of a four-speed transmission.
136
6293_Book.indb 136 1/11/13 3:54 PM

Traditional shift maps are throttle-position based, and are widely used in the throttle-position-
based powertrain control architectures. In the newly developed power-based powertrain
control architectures, the shift maps have evolved into power-based.
Designs of the shift map of a motor vehicle are critical for achieving a good balance of perfor-
mance, drive quality, and fuel economy. Various methods and techniques have been developed
by the automatic transmission calibration community throughout the years; many of them are
mainly trial-and-error methods. With these traditional methods, the optimality of the result-
ing shift map is unclear. To change this situation, methods of using mathematical tools to
generate shift maps have been developed. In the next section, a dynamic programming method
for generating throttle-position-based shift maps is presented. In a later section, a different
method for generating power-based shift maps will be presented when the topic of integrated
powertrain controls is introduced.
3.6.2 Dynamic-programming-based shift map

generation
Dynamic programming is a global optimization technique that is used to solve multi-stage
decision-making and control problems. This section introduces the concept of using dynamic
programming methods to generate a throttle-position-based shift map for an automatic trans-
mission equipped motor vehicle.
First, let’s take a brief look at what dynamic programming optimization is. The basic idea of
dynamic programming is described by the principle of optimality stated by Bellman [3-6]:
“An optimal policy has the property that, whatever the initial state and optimal first
decision may be, the remaining decisions constitute an optimal policy with regard to
the state resulting from the first decision.”
To obtain a better understanding of the previous statement, let’s consider an optimization
problem of the following: minimize or maximize the cost function shown in Eq. 3.79.
N −1
J = G N ( x (N )) + ∑ Lk ( x (k ), u(k ), w (k )) (3.79)
k =0
Here,
x (k + 1) = f ( x (k ), u(k ), w (k )) , k = 0,1,…, N − 1 (3.80)
Equation 3.80 is the system equation, and is subject to Eq. 3.81:
x ∈ X (k ) ⊂ ℜn , u ∈ U ( x (k ), k ) ⊂ ℜm (3.81)
Here, x(k) is the state variable at stage k in the space of X(k), u(k) is the control variable, w(k) is
a predetermined disturbance, f is the transition function that represents the system model, Lk
is the instantaneous transition cost at stage k, and GN is the cost at final stage N.
The principle of optimality states that if U = {u0, u1, u2, …, uN–1}, where uk maps states x(k) into
controls uoptimal(k) = uk(x(k)), is the optimal control strategy that minimizes (maximizes) the
137
6293_Book.indb 137 1/11/13 3:54 PM

Chapter 3
given cost function, then the truncated policy Ũ = {ui, ui+1, ui+2, …, uN–1} is the optimal control
strategy for the subproblem of minimizing (maximizing) the cost function in Eq. 3.82:
N −1
J = G N ( x (N )) + ∑ Lk ( x (k ), u(k ), w (k ))
k =i
(3.82)
From this fact, it is quite intuitive to tackle the problem starting from the last stage. The
optimal policy can be obtained if we first solve a one-stage subproblem involving only the last
stage and then gradually extend to subproblems involving the last two stages, last three stage,
etc., until the entire problem is solved. In this manner, the overall optimization problem can be
decomposed into a sequence of simpler minimization problems, as follows.
Step N – 1:
J ∗ ( x (N − 1)) = min {G N ( x (N )) + LN −1 ( x (N − 1), u(N − 1), w (N − 1))} (3.83)

u ( N −1)
Step k, for 0 ≤ k < N – 1:
J ∗ ( x (k )) = min { J * ( x (k + 1)) + Lk ( x (k ), u(k ), w (k ))} (3.84)

u (k )
Where J ∗ ( x (k )) is the optimal cost-to-go function or optimal value function at state x(k),
starting from time stage k. It represents the optimal resulting cost if at stage k the system starts
at state x(k) and follows the optimal policy thereafter until the final stage.
The backward-searching strategy based on the principle of optimality can be simplistically
and clearly illustrated using the shortest-path problem [3-7]. The problem is illustrated in
Fig. 3.90. The objective is to find the nearest coast city among three coast cities (N3, C3, and
S3), with the starting point of inland city C0. There are two intermediate stages the traveler will
go through. In each stage, the traveler has three intermediate cities from which to choose. The
cities N1, C1, and S1 are the first-stage cities from which to choose, and the cities N2, C2, and
S2 are the second stage cities. The distance between two cities is labeled next to the line con-
necting the cities.
One way to solve the problem is by forward searching. Starting from city C0, the traveler
simply finds the shortest path that will lead from C0 to one of the first-stage cities, and it is
C0-C1. Then, the traveler finds the shortest path that will lead from C1 to one of the sec-
ond-stage cities, and it is C1-C2. By repeating this procedure, the traveler obtains a path of
C0-C1-C2-C3, with a total distance of 1550, as shown in Fig. 3.91. Is this the shortest path?
There is no guarantee, because not all the possible paths are considered. Because this forward-
searching method only pursues the instant best interests at each stage, it is also called the
greedy method.
The backward-searching strategy based on the principle of optimality guarantees the consider-
ation of all possible paths, and therefore provides a global optimization solution. The method
starts from the last stage. First, the method finds the shortest path from each of the second-
stage cities (N2, C2, S2) to the nearest coast city. The result is shown in Fig. 3.92. The shortest
distance from each of the second-stage cities (N2, C2, S2) to the nearest coast city is labeled
next to the city, and the shortest path is highlighted with thicker lines.
138
6293_Book.indb 138 1/11/13 3:54 PM

Fig. 3.90 Graphic

illustration of shortest-
path problem.
Fig. 3.91 The forward-

searching result of the
shortest-path problem.
Fig. 3.92 First step of

backward searching: solve
the last (third) stage of the
shortest-path problem.
139
6293_Book.indb 139 1/11/13 3:54 PM

Chapter 3
Then the method goes one stage backward to find the shortest path from each of the first-stage
cites (N1, C1, S1) to the nearest coast city. To find the shortest path from N1 to the nearest
coast city, using the method based on the principle of optimality, simply add the distance from
N1 to each of the second-stage cities (N2, C2, S2) to the number next to the city, and choose
the one with the shortest distance, as illustrated in Fig. 3.93. The result of this step is shown
in Fig. 3.94. The number next to each city is the shortest distance from this city to the nearest
coast city; the shortest path is highlighted in thicker lines.
Then the method goes another stage backward to finally find the shortest path from the
starting city C0 to the nearest coast city. To do this using the method based on the principle of
optimality, simply add the distance from Co to each of the first-stage cities (N1, C1, S1) to the
number next to the city, and choose the one with the shortest distance. The result is C0-S1-S2-
S3 with a distance of 1400, as shown in Fig. 3.95.
Fig. 3.93 The second step

of backward searching:
solve the second stage of
the shortest-path problem.
Fig. 3.94 Result of

the second step of
backward searching.
140
6293_Book.indb 140 1/11/13 3:54 PM

Fig. 3.95 Result of the

third step of backward
searching.
The shortest path found using the backward-searching method is shorter than the one found
using the forward-searching method. Because the backward-searching method practically
considered all potential paths, the result is the true shortest path.
Now let’s consider how to use the dynamic programming method to generate a throttle-based
shift map for an automatic transmission. The problem is formulated as shown in Fig. 3.96. The
system stats are vehicle speed (V), engine speed (Ne), and gear position (Gs). The controls are
gear up, down, or hold. The throttle opening is considered as a disturbance to the system. The
transition costs are fuel consumption and distance.
The dynamic programming optimization of shift maps is formulated as a constrained optimi-
zation problem [3-8, 3-9]. The optimal shift point determination is performed at each throttle
opening level. The cost to be minimized is the “fuel per distance” of the result of driving the
vehicle, starting from zero speed at the throttle opening for a certain amount of time. The
constraint is a vehicle speed profile that specifies the expected performance for that throttle
position. The constraint demands that when driving the vehicle at the throttle opening level,
starting from zero speed, the vehicle speed always has to be above the specified speed profile.
The vehicle speed profile constraint lines are generated by lowering the best-performance
vehicle speed lines for certain percentages, depending on the throttle level. For lower throttle
Fig. 3.96 Dynamic

programming formulation for
shift map generation.
141
6293_Book.indb 141 1/11/13 3:54 PM

Chapter 3
levels the speed lines are lowered a large percentage; for higher throttle levels they are lowered
a smaller percentage; for near 100% throttle level they are lowered zero percentage. The
best-performance vehicle speed lines are the vehicle speed lines when shifting at the best-per-
formance shift points of each throttle level. The best-performance shift points are the vehicle
speeds of the intercept points of vehicle traction effort of each gear for each throttle level, as
illustrated in Fig. 3.89.
Figure 3.97 illustrates this concept of constrained shift-point optimization, using the case of
20% throttle position as an example. In Fig. 3.97, the horizontal axis is the time step, and the
vertical axis is the vehicle speed. The numbers in the cells are gear positions. The lines con-
necting the numbers in the cells represent control actions of up-shift, down-shift, or holding
the current gear. The numbers on the lines are fuel and distance of the result of the control
action. The dotted line is the vehicle speed constraint. The fuel and distance associated with a
control decision (gear shifting up, down, or hold) at each state is calculated using a dynamic
simulation model. After this table (can be considered as a road map) is created, the backward-
searching method can be used to find the optimal path that provides the minimum total
“fuel/distance,” which is always above the vehicle speed constraint. The optimal shift points
for the throttle opening can be obtained from this optimal path. The throttle-based shift map
is generated after the previous optimal shift point calculations are performed at all throttle
opening levels.
3.6.3 Artificial intelligence-based shift scheduling

system
Various Artificial Intelligence (AI)-based shift scheduling systems have been developed to
make gear-shifting speeds adaptive to vehicle loads, road conditions, and other factors. In
this section, an AI-based shift scheduling system shown in Fig. 3.98 will be introduced. In
this system, the up-shift and downshift points are determined by a fuzzy logic algorithm.
The input information to the fuzzy logic algorithm can be as many as required; for example,
Fig. 3.97 Shift point search for 20% throttle opening.
142
6293_Book.indb 142 1/11/13 3:54 PM

Fig. 3.98 AI-based shift scheduling system.
throttle position, vehicle acceleration, steering position, GPS information, vehicle weight, road
grade, etc. Since more information is taken into consideration for determining shift points,
the AI-based shift scheduling system is more adaptive to the driving conditions and driver’s
desires.
In the following sections, some basic concepts of fuzzy logic are introduced first, and then the
fuzzy logic-based shift scheduling system is presented.
3.6.3.1 Fuzzy set, membership function, and truth value

In the traditional logic theory, the truth value of a statement is represented only as 0 or 1. If the
statement is true, then the truth value of the statement is 1; if the statement is false, then the
truth value of the statement is 0. This is illustrated in the upper portion of Fig. 3.99. Here, four
Fig. 3.99 Definitions of fuzzy sets and their membership functions.
143
6293_Book.indb 143 1/11/13 3:54 PM

Chapter 3
conventional sets, Cold, Cool, Warm, and Hot, are defined in the relevant temperature range.
The membership functions for those conventional sets have a shape of rectangular, so the truth
value of those membership functions only could be 0 or 1. For the current temperature indi-
cated by the long vertical line in the figure, the truth values of the statements: “temperature is
cold,” “temperature is cool,” “temperature is warm,” and “temperature is hot” are shown in the
table next to the conventional sets drawing.
In the fuzzy logic theory [3-10, 3-11], the membership functions of fuzzy sets are nonrect-
angular; for example, they could be triangular as shown in the lower portion of Fig. 3.99.
Therefore, the truth value of a statement could be between 0 and 1. For the current tem-
perature indicated by the long vertical line in the figure, the truth values of the statements:
“temperature is cold,” “temperature is cool,” “temperature is warm,” and “temperature is hot,”
based on the illustrated membership function of the fuzzy sets, are given in the table next to
the fuzzy sets drawing.
3.6.3.2 Fuzzy logic operations

The basic fuzzy logic operations are “and,” “or,” and “not.” The result of the “and” operation on
two statements is the minimum truth value of the two statements. The following is an example
of an “and” operation.
• And: minimum
Truth value of {T is Low and DT is Mid} = min (0.7, 0.2) = 0.2. Here, as shown in Fig. 3.100,
the truth value of the statement {T is Low} is 0.7, and the truth value of the statement {DT is
Mid} is 0.2.
The result of the “or” operation on two statements is the maximum truth value of the two state-
ments. The following is an example of an “or” operation.
• Or: maximum
Truth value of {T is Low or DT is Mid} = max (0.7, 0.2) = 0.7.
The result of the “not” operation is the complement truth value of the statement. The following
is an example of a “not” operation.
• Not: complement
Truth valve of {not T is low} = 1 – 0.7 = 0.3
Fig. 3.100 Member shift function and truth value of T is high and DT is Mid.
144
6293_Book.indb 144 1/11/13 3:54 PM

3.6.3.3 Fuzzy controls

Let’s use the oven temperature control as an example to explain fuzzy controls. The block
diagram of such control systems is shown in Fig. 3.101.
The input variables to the fuzzy controller are the temperature error, Terror, and the change
of temperature error, DTerror. The control variable from the fuzzy controller is the heater
current change, ∆Iheater. The desired heater current can be calculated from the current heater
current and ∆Iheater. The membership functions defined for those three variables are shown in
Fig. 3.102. The names of those membership functions are defined as Z: zero, N: negative, P:
positive, NS: negative small, NL: negative large, PS: positive small, and PL: positive large.
The fuzzy controller is defined by the following fuzzy control rules.
1. If Terror is Z and DTerror is Z, then ∆Iheater is Z.
2. If Terror is NS and DTerror is N, then ∆Iheater is PL.
3. If Terror is PL and DTerror is Z, then ∆Iheater is NS.
4. If Terror is NS and DTerror is Z, then ∆Iheater is PS.
5. If Terror is PL and DTerror is P, then ∆Iheater is NL.
The fuzzy control rules can be as many as needed. Here, we assume we only need these five
rules as an example. The statements following “If ” are the conditions, and the statements
Fig. 3.101 Block diagram of fuzzy control system of oven temperature.
Fig. 3.102 Membership functions defined for Terror, DTerror, and ∆Iheater.
145
6293_Book.indb 145 1/11/13 3:54 PM

Chapter 3
following “then” are the control actions. To generate the control signal from these fuzzy control
rules, first each rule is evaluated as illustrated in Figs. 3.103a to 3.103e. In the figures, the value
of the actual temperature error and its change are represented as Ťerror and DŤerror, respectively.
Figure 3.103a evaluates rule 1: {If Terror is Z and DTerror is Z, then ∆Iheater is Z}. As shown in the
figure, the truth value of the first statement {Terror is Z} is 0.3; the truth value of the second
statement {DTerror is Z} is 0.5. Because it is an “and” operation of these two statements, the
resulting truth value is 0.3. Using this truth value and based on the control action statement
{then ∆Iheater is Z}, an isosceles trapezoid area is obtained from the membership function of
Zero ∆Iheater, with height of 0.3 as shown in the figure. The other control rules are evaluated in
the same manner.
Fig. 3.103a Evaluation of rule 1: If Terror is Z and DTerror is Z, then ∆Iheater is Z.
Fig. 3.103b Evaluation of rule 2: If Terror is NS and DTerror is N, then ∆Iheater is PL.
146
6293_Book.indb 146 1/11/13 3:54 PM

Fig. 3.103c Evaluation of rule 3: If Terror is PL and DTerror is Z, then ∆Iheater is NS.
Fig. 3.103d Evaluation of rule 4: If Terror is NS and DTerror is Z, then ∆Iheater is PS.
Fig. 3.103e Evaluation of rule 5: If Terror is PL and DTerror is P, then ∆Iheater is NL.
147
6293_Book.indb 147 1/11/13 3:54 PM

Chapter 3
The areas obtained from evaluating all fuzzy rules are combined together as shown in Fig.
3.104. The value of the control variable ∆Iheater then is obtained through defuzzification. One
common defuzzification method is the weighted average method. The resulting value of the
control variable ∆Iheater obtained from the weighted average method is shown in Fig. 3.104 as
∆Ĭheater. As suggested by the name, ∆Ĭheater is the point that divides the combined area into two
parts with equal weight.
3.6.3.4 Fuzzy logic rules for up-shift and down-shift points

In the AI-based shift schedule system shown in Fig. 3.98, the up- and down-shift points are
generated by a fuzzy logic algorithm. Now with the understanding of basic concepts of fuzzy
logic, we are ready to discuss the fuzzy logic algorithm for shift point determination [3-12,
3-13]. Assume that the input variables to the algorithm are throttle opening and vehicle
acceleration. The membership functions defined for these two input variables are shown in
Fig. 3.105. The upper graph defines four membership functions for the variable of
throttle opening. They are ThZ(throttle zero), ThS(throttle small), ThL(throttle large), and
ThVL(throttle very large). The lower graph defines five membership functions for the variable
of vehicle acceleration. They are AcNL(acceleration negative large), AcNS(acceleration negative
small), AcZ(acceleration zero), AcPS(acceleration positive small), and AcPL(acceleration
positive large).
Fig. 3.104 Defuzzification to obtain the value of control variable ∆Iheater.
Fig. 3.105 Membership function of throttle opening and vehicle acceleration.
148
6293_Book.indb 148 1/11/13 3:54 PM

The fuzzy logic rules for the up-shift and down-shift point determination are defined as
follows.
Rules for determining up-shift point:

• If Th is S and Ac is PL, then up-shift point is Low.
• If Th is S and Ac is PS, then up-shift point is Low.
• If Th is L and Ac is PL, then up-shift point is Mid.
• If Th is L and AC is PS, then up-shift point is High.
Rules for determining down-shift point:

• If Th is ZO and Ac is NS, then down-shift point is Low.
• If Th is ZO and Ac is NL, then down-shift point is High.
• If Th is VL and Ac is ZO, then down-shift point is High.
• If Th is VL and Ac is NS, then down-shift point is Very High.
• If Th is VL and Ac is NL, then down-shift point is Very High.
Define six constants as:

• lowusp: low up-shift point.
• midusp: middle up-shift point.
• hiusp: high up-shift point.
• lowdsp: low down-shift point.
• hidsp: high down-shift point.
• vhidsp: very high down-shift point.
Then the up-shift point wu and down-shift point wd can be calculated using the weighted
average method shown in Eqs. 3.85 and 3.86.
λlow lowusp + λmid midusp + λhi hiusp

ωu = (3.85)
λlow + λmid + λhi
βlow lowdsp + βhi hidsp + βvhi vhidsp

ωd = (3.86)
βlow + βhi + βvhi
Here,
llow = max(min(ThS(th), AcPL(ac), min(ThS(th), AcPS(ac))), obtained from evaluating the

first two rules for up-shift point;
lmid = min(ThL(th), AcPL(ac)), obtained from evaluating the third rule for up-shift point;
lhi = min(ThL(th), AcPS(ac)), obtained from evaluating the fourth rule for up-shift point;
149
6293_Book.indb 149 1/11/13 3:54 PM

Chapter 3
blow = min(ThZO(th), AcNS(ac)), obtained from evaluating the first rule for down-shift point;
bhi = max(min(ThZO(th), AcNL(ac)), min(ThVL(th), AcZO(ac))), obtained from evaluating
the second and third rules for down-shift point;
bvhi = max(min(ThVL(th), AcNS(ac)), min(ThVL(th), AcNL(ac))), obtained from evaluating
the last two rules for down-shift point.
The shift speed fuzzy logic algorithm just described not only uses the throttle opening level as
an input variable, but also uses vehicle acceleration as an input variable. Because the vehicle
acceleration is determined by the factors of vehicle mass, vehicle load, road grade, etc., the shift
speeds calculated from the fuzzy logic algorithm are adaptive to those factors.
3.7 Integrated powertrain controls for

driveability and fuel economy
Gear-shifting schedule control systems described in the previous sections all use the throttle
opening as the input variable to determine up- or down-shift speeds. This is because before
the electronically controlled throttle became available, the throttle was directly controlled by
drivers through mechanical linkages. Since the electronic throttle became available, integrated
powertrain control, namely integrated control of engine operation and gear selection, has
become possible. The integrated powertrain control provides an opportunity to achieve total
optimal operation of powertrain systems, and to improve both fuel economy and drive quality
[3-14, 3-15, 3-16].
The main topics of integrated powertrain controls include: (1) How to interpret the driver’s
pedal input signal. Should it be interpreted as a demand for throttle, torque, or power? (2)
How to control the engine. Should it be throttle-based or engine-torque-based? (3) How to
determine gear shift speed. Should it remain throttle based, or should it be based on some
other parameters such as power or torque? (4) How to control gear shifts in such an integrated
powertrain control environment.
In the following sections, an integrated powertrain control system will be described. The
advantages of this integrated powertrain control system are: additional degrees of freedom to
optimize powertrain operations; improvement of consistency and accuracy of drive torque
control; reduction of gear-shifting busyness; allowing more-fuel-efficient shift patterns without
losing drivability; automatically compensating for engine torque variation due to altitude
changes and engine hardware variations, etc; maintaining output torque unchanged before,
during, and after a power-on up-shift; separation of the application-specific subjective tuning
and math-based objective optimization; and finally, simple computations.
3.7.1 Overall architecture of integrated powertrain

control system
The overall architecture of the integrated powertrain control system is shown in Fig. 3.106.
The system includes a pedal input interpreter, an engine torque command generator, an engine
torque control system, a power-based transmission gear selection system, and a gear-shift
control algorithm.
The pedal input interpreter interprets the driver’s pedal input signal (pedal position sensor—
PPS) as an engine power demand. A look-up table is used to implement such interpreters. The
150
6293_Book.indb 150 1/11/13 3:54 PM

Fig. 3.106 Overall architecture of the integrated powertrain control system.
independent variables of the look-up table are the pedal position and vehicle speed. The output
variable of the look-up table is the driver’s engine power demand. The typical shape of the
look-up table is shown in Fig. 3.107. For a given pedal position, the engine power demand goes
up as the vehicle speed increases. Such driver’s input interpreters provide a natural driving
experience. The calibration of the pedal input interpreter determines the powertrain response
to the driver’s demand, and therefore is the major control component to deliver the desired
driving characteristic for a specific vehicle application. The interpreter could be adaptive to
other parameters such as vehicle load, road grade, etc.
The engine torque command generator generates the engine torque command, and sends
it to the engine torque control system. The engine torque command is generated simply by
dividing the engine power demand by the current engine speed, and limited by the desired
launch torque at low vehicle speed. The desired launch torque is a function of engine power
demand. It is clear that after a power-on up-shift, the engine torque command will increase
due to the decrease of the engine speed to maintain the desired engine power. Therefore, after
a constant-pedal-position power-on up-shift, the transmission output torque will maintain the
same level as before the shift. This greatly improves the vehicle drivability in comparison with
the traditional throttle-based powertrain control system, in which drivers will feel a decrease
of traction force after a power-on up-shift.
The engine torque control system controls various engine control parameters to achieve
the desired engine torque using a feedback/feed-forward control algorithm. The high-level
diagram of such control systems is shown in Fig. 3.108. The feedback control compensates
Fig. 3.107 Pedal input

interpreter.
151
6293_Book.indb 151 1/11/13 3:54 PM

Chapter 3
Fig. 3.108 High-level diagram of engine torque control system.
for the variations of altitude, engine hardware, etc. to ensure the accuracy of engine
torque controls.
The power-based transmission gear selection algorithm selects transmission gears based on
the power demand and vehicle speed. The detailed descriptions of such algorithms are pre-
sented in Section 3.7.2.
The gear-shift control algorithm not only controls clutch pressures but also controls engine
torque during a gear shift to achieve smooth gear shifts. As an example, the constant-output-
torque power-on up-shift controls enabled by this integrated powertrain control system will be
presented in Section 3.7.3.
3.7.2 Power-based gear selection

In the integrated powertrain control system, the transmission gears are selected based on the
engine power demand. Two algorithms for power-based gear selection are described in the fol-
lowing paragraphs. One uses the power-based shift maps to determine the desired gear ratio.
Another one is a real-time gear selection algorithm that evaluates the fuel efficiency of each
gear in real time and selects gears to maximize fuel economy under drivability constraints.
A power-based shift map is shown in Fig. 3.109. In the power-based shift map, the up- and
down-shift speeds are functions of engine power demand, not throttle as in the traditional
powertrain control system. The solid lines labeled 1-2, 2-3, etc. are up-shift lines. The dashed
lines labeled 2-1, 3-2, etc. are down-shift lines.
Various computer-aided tools have been developed for generating power-based shift maps. The
basic concept of these tools is to evaluate all shift points and find a shift map that maximizes
fuel economy under drivability constraints.
The power-based shift map generation tool could be formulated as a constrained optimization
problem. The cost function, J, is defined as the inverse of the fuel efficiency, as shown in
Eq. 3.87. The drivability constraints include torque reserves as functions of power level,
minimum and maximum engine and turbine speeds as functions of power level, TCC apply
conditions, hysteresis between up and down shifts, dead pedal travels, etc.
engine _ fuel _ rate

J= (3.87)
axle _ power
152
6293_Book.indb 152 1/11/13 3:54 PM

Fig. 3.109 Power-based shift map.
Figure 3.110 illustrates the basic concept of power-based gear selection. Assume that the trans-
mission is a six-speed transmission. The engine speeds of all six gears can be calculated for the
current vehicle speed. Then they can be placed on the equate power line on the engine torque-
speed plot. The 1st gear over the maximum speed limit therefore is eliminated. The 5th and
6th gears are also eliminated due to the violation of torque reserve constraint. Then the desired
gear is chosen within 2nd, 3rd, and 4th gears to minimize the defined cost function.
The transmission gear selection method just described could be used to program a com-
puter-aided tool to generate a power-based shift map, and also could be programmed in
transmission controllers to form a real-time gear-selection algorithm. Figure 3.111 shows a
real-time transmission gear selection algorithm. The algorithm calculates engine speed, engine
torque, turbine speed, and turbine torque of each gear based on the desired engine power and
current vehicle speed. The results of this calculation are then used by the optimal gear selection
algorithm to determine the desired gear. Various cost functions could be used for the optimal
Fig. 3.110 Power-based transmission gear selection.
153
6293_Book.indb 153 1/11/13 3:54 PM

Chapter 3
Fig. 3.111 Diagram of real-time gear-selection algorithm.
gear selection based on applications. For example, the cost function defined in Eq. 3.87 could
be the cost function for gasoline engines, and the weighted function of BSFC (Brake-Specific
Fuel Consumption) and emission could be used in the case of diesel engines.
It should be noted that the power-based transmission gear selection algorithm is independent
from the pedal input interpreter. Changes of calibrations of the pedal input interpreter do
not require changes of calibrations of gear-selection algorithms. The pedal input interpreter
provides the desired response to the driver’s demand, and the power-based gear selection algo-
rithm determines the optimality of the powertrain operations.
3.7.3 Constant-output-torque power-on up-shift control

In the traditional throttle-based powertrain control system, after a power-on up-shift, because
the engine power decreases as a result of engine speed decrease, the transmission output
torque will decrease to a lower level. In the power-based integrated powertrain control system,
because the engine power demand remains the same, the transmission output torque main-
tains nearly unchanged before and after a power-on up-shift. Now, the question is, can the
transmission output torque also remain unchanged during a power-on up-shift? The answer is
yes. A clutch-to-clutch shift-control algorithm is developed that keeps the transmission output
torque unchanged, even during a power-on up-shift.
The traditional and the constant-output-torque power-on up-shift controls are shown in Figs.
3.112 and 3.113, respectively.
Figure 3.112 shows the traditional clutch-to-clutch power-on up-shift control. In the tra-
ditional control, there is no active control of the engine torque during the torque phase. So
the transmission input torque is maintained at the same level before the shift, during the
torque phase of the shift, and after the shift. Therefore, the output torque decreases to a lower
level after the gear shift. In the inertial phase, the input speed change is achieved by engine
torque reduction.
Figure 3.113 shows the constant-output-torque clutch-to-clutch power-on up-shift control. In
this control, the transmission input torque is actively controlled during the torque phase. As
the oncoming clutch (high gear clutch) torque is commanded to ramp up during the torque
phase, the transmission input torque is also commanded to ramp up based on Eq. 3.88.
154
6293_Book.indb 154 1/11/13 3:54 PM

rl − rh
Tin = Tin _ initial + Thc (3.88)
rl ⋅ K hc
Here, Tin_initial is the input torque at the beginning of the shift, Thc is the high gear clutch
(oncoming clutch) torque command, Tin is the input torque command, rl is the low gear ratio,
rh is the high gear ratio, and Khc is the input torque ratio of the high gear clutch. The end value
of input torque ramp-up command is given in Eq. 3.89.
rl 1
Tin _ end = Tin _ initial = Tout _ initial (3.89)
rh rh
Therefore, the end value of the oncoming clutch command should be as given in Eq. 3.90.
rl ⋅ K hc
Thc _ end = Tin _ initial (3.90)
rh
Fig. 3.112 Traditional clutch-to-clutch power-on up-shift control.
Fig. 3.113 Constant-output-torque clutch-to-clutch power-on up-shift control.
155
6293_Book.indb 155 1/11/13 3:54 PM

Chapter 3
The end value of the oncoming clutch command defined in Eq. 3.90 is also called the inertial
phase torque. The oncoming clutch torque command is maintained at this level during the
inertial phase. The off-going clutch (low-gear clutch) torque command has to be reduced to
zero when the oncoming clutch torque is ramped up to this end value. In the inertial phase, the
input speed change is also achieved by the input torque reduction from the new level given in
Eq. 3.89. At the end of the inertial phase, the input torque is recovered to the same new level.
This new level of the input torque is the torque required to maintain the engine power at the
same level after the shift. As a result of this control strategy, the transmission output torque is
maintained at the same level before, during, and after the gear shift.
Figure 3.114 compares vehicle test data of block pedal vehicle acceleration traces of power-
based and throttle-based powertrain controls. It can be seen that the power-based powertrain
control provides a much smoother vehicle acceleration and gear shift.
3.8 Centrifugal pendulum vibration absorber

Having the ability to lock up torque converter clutches at low engine speed becomes not just
desirable but indispensable to further improve fuel economy of automobiles. Traditional
spring-inertia-based torsional dampers have limitations in achieving such an objective. These
limitations have become more pronounced, as many new engine technologies adapted for
improving fuel efficiency and performance generate higher magnitudes of engine torque fluc-
tuations. In such cases, a Centrifugal Pendulum Vibration Absorber (CPVA) is a potential
technology to enable torque converter clutch lockup at lower engine speeds.
The unique characteristic of a CPVA is that the oscillation frequency of a centrifugal pendulum
is proportional to the angular speed of the shaft to which it is attached. Therefore, the centrifu-
gal pendulum of a CPVA can be designed to oscillate the same number of times per revolution
as the combustion event of an internal combustion engine, such that it generates a counter-
acting torque opposing the oscillatory engine torque; therefore, it reduces net engine torque
fluctuations. For example, for a six-cylinder engine, a CPVA is designed to oscillate three times
per revolution; for a four-cylinder engine, the CPVA is designed to oscillate two times per
revolution. This unique speed-adaptive characteristic of the absorption frequency of CPVAs
Fig. 3.114 Comparison of acceleration traces for a vehicle with power-based and throttle-based powertrain controls.
156
6293_Book.indb 156 1/11/13 3:54 PM

is in contrast to the traditional inertia-spring torsional absorbers that have a fixed absorption
frequency. Therefore, CPVAs have unique advantages for reducing torsional vibrations induced
by an internal combustion engine.
In this section, the basic concept and designs of CPVAs will be introduced first. Then the equa-
tions of motion of CPVAs will be derived. Finally, design issues will be discussed.
3.8.1 Basic concept and various designs for CPVA

Figure 3.115 shows the basic concept of a CPVA. A number of centrifugal pendulums are
attached to the main shaft, which is driven by an oscillatory engine torque. The pendulums
oscillate under the restoring effort of the centrifugal force as the main shaft rotates. If the
CPVA is designed correctly, the centrifugal pendulums oscillate the same number of times
per engine revolution as the engine combustion events, but in the opposite direction of engine
torque impulse. As a result, opposing torque is generated to cancel out or reduce the engine
torque fluctuations. Figure 3.116 illustrates such engine torque impulse cancelations.
Figures 3.117–3.119 present some alternative designs of CPVAs. Figure 3.117 shows the bifilar
centrifugal pendulum. It can be shown that the center of mass (CM) of a bifilar centrifugal
pendulum makes a circular motion, as does the basic centrifugal pendulum shown in Fig. 3.115.
Figure 3.118 illustrates the roller-type centrifugal pendulum. The roller pendulum oscil-
lates inside the circular track. The path of the CM of a roller centrifugal pendulum is also a
circular arc.
Figure 3.119 presents an alternate design of bifilar centrifugal pendulums. The pins are
implanted on the main shaft. The tracks opened on the pendulum block guide the motion of
Fig. 3.115 Conceptual diagram of centrifugal

pendulum vibration absorber.
Fig. 3.116 Engine torque fluctuation is

reduced by the pendulum torque.
157
6293_Book.indb 157 1/11/13 3:54 PM

Chapter 3
the pendulum. The shape of the track determines the path of the CM of the pendulum. There-
fore, with this design, the path of the pendulum motion can be designed to be any desired shape.
Figure 3.120 is another alternate design of bifilar centrifugal pendulums. In this design, track
A is opened on the main shaft, and track B is opened on the pendulum. Track A and track
B both ride on the roller at the opposite side. This design also allows any desired shape of
pendulum path.
Fig. 3.117 Bifilar centrifugal pendulum.
Fig. 3.118 Roller centrifugal pendulum.
Fig. 3.119 An alternate design of

bifilar centrifugal pendulums.
Fig. 3.120 Another design of

bifilar centrifugal pendulums.
158
6293_Book.indb 158 1/11/13 3:54 PM

3.8.2 Equations of motion of CPVA

This section derives equations of motion of a CPVA. For that, let’s establish the coordinate
system shown in Fig. 3.121 to describe the motion of the CM of the pendulum. Here, the point
o is the center of the main shift; the point p is the center of the curvature of the pendulum path
at f = 0; R is the distance from o to p; m is the mass of the pendulum; l is the distance from
the point p to the center of mass of the pendulum; q is the angular displacement of the main
shaft; f is the pendulum swing angle (f = 0 means the CM of the pendulum on the extension
line of o to p); b is the radius of curvature of the pendulum path at f = 0. If l is a constant, the
pendulum path is of a circular arc. Because we would like to model general paths of pendulum
motion, the l is expressed as a function of f, as shown in Eq. 3.91.
l = f (φ ) (3.91)
With the presented coordinate system, the Lagrange function of the system is obtained as
shown in Eq. 3.92. Here, the potential energy of the pendulum mass due to the gravity force
is ignored.
1 2 1 1 1
L= Jθ + mR 2θ 2 + ml2 + ml 2 (θ + φ)2 + mRθl sin(φ )
2 2 2 2 (3.92)

+mRl cos(φ )θ + mRl cos(φ )θφ
2
Based on the Lagrange mechanics, the equations of motion of the CPVA are obtained using the
Lagrange formulas of Eqs. 3.93 and 3.94. The resulting equations of motion of the CPVA are
shown in Eqs. 3.95 and 3.96.
d  ∂L  ∂L
 − =T (3.93)
dt  ∂θ  ∂θ
d  ∂ L  ∂L
 − =0 (3.94)
dt  ∂φ  ∂φ
Fig. 3.121 A coordinate system to describe

the motion of the pendulum mass.
159
6293_Book.indb 159 1/11/13 3:54 PM

Chapter 3
 dl 
( J + mR 2
+ ml 2 + 2mRl cos(φ ))θ+ ml 2 + mRl cos(φ ) + mR sin(φ )  φ = T +
 dφ 
 dl d 2l dl 
mRl sin(φ ) − 2ml − mR sin(φ ) 2 − 2mR cos(φ ) φ 2 + (3.95)
 dφ dφ dφ 
 dl dl 
2mRl sin(φ ) − 2mR cos(φ ) − 2ml θφ
 dφ dφ 
2   2  dl  
2
dl
l + R sin(φ ) + Rl cos(φ )θ + l +    φ =
 dφ    dφ  
(3.96)
 dl dl   dl dl d 2l  2
l + R cos(φ ) − Rl sin(φ )θ 2 − l +  φ
 dφ dφ  dφ dφ dφ 
2

Here, all variables in Eqs. 3.95 and 3.96 are defined in Fig. 3.121, other than J. J is the inertia
of the main shaft. The equations of motion of CPVAs have two degrees of freedom, and are
coupled and nonlinear.
Let’s assume that the main shaft rotates at a constant speed Ω; that is, θ = 0 and θ = Ω .eps.
Let’s also assume that f is very small; that is, sin(f) = f and dl/df = 0. Then Eq. 3.96 becomes
Eq. 3.97.
R
φ + Ω2φ = 0 (3.97)
b
From Eq. 3.97, we conclude that under the assumed conditions, the natural frequency wn of
the pendulum can be expressed as shown in Eq. 3.98.
ωn = Ω R / b = Ωn (3.98)
Here, R and b are defined in Fig. 3.121; n = R / b is called the order of the CPVA, which
defines the number of the pendulum oscillation per revolution. If n is equal to the number of
engine combustion per revolution, then the CPVA is perfectly tuned. However, because the
actual system is nonlinear, and the pendulum oscillates in a larger f region to obtain meaning-
ful vibration absorption, the tuning of CPVAs has to be performed using simulation models,
not the previous formula.
Next let’s consider, mathematically, why a CPVA can reduce torsional vibration. First let’s
assume that the engine torque is expressed as shown in Eq. 3.99.
Te = T0 + A1 sin(ωnt ) (3.99)
If the pendulum is perfectly tuned, then the motion of the pendulum under the excitation
torque of Eq. 3.99 can be expressed as Eq. 3.100.
φ = φa sin(−ωnt ) (3.100)
160
6293_Book.indb 160 1/11/13 3:54 PM

Here, fa is the magnitude of the pendulum oscillation. If we also assume f is small, that is,
sin(f) = f, cos(f) = 1, dl/df = 0; then Eq. 3.95 could be rewritten as shown in Eq. 3.101.
( J + mR φ )θ = T + {A − (mRl + mR )θ φ

2 2
0 1
2 2
a − mRlφ 2φa } sin(ωnt ) + 2mRlφφθ
(3.101)
What Eq. 3.101 tells us is that the torque fluctuation exerted on the main shaft has been
reduced from A1 to { A1 − (mRl + mR 2 )θ 2φa − mRlφ 2φa } . This is why CPVAs can reduce tor-
sional vibrations.
3.8.3 Simulink model of CPVA

To build a Simulink simulation model for CPVAs, Eqs. 3.95 and 3.96 are arranged into a
matrix format, as shown in Eq. 3.102.
 
θ 
M =B (3.102)
 φ 
Here, the definition of matrix M and B could be obtained from Eqs. 3.95 and 3.96. Readers are
encouraged to go through this exercise.
The Simulink model based on the matrix format of the equations of motion of CPVAs is given
in Fig. 3.122. All elements in Eq. 3.102 are labeled on the model.
Fig. 3.122 Simulink simulation model of CPVA.
161
6293_Book.indb 161 1/11/13 3:54 PM

Chapter 3
3.8.4 Tuning of CPVA

As mentioned earlier, the goal of tuning a CPVA is to find the correct value of R and b such
that the pendulum oscillates the same number of times as the engine combustion for each
engine revolution. That will make the pendulum order n equation the engine order. Although
Eq. 3.98 provides an expression of pendulum order n as a function of parameters R and b, the
actual tuning has to be performed using the nonlinear simulation model just developed. This
is because Eq. 3.98 is obtained based on the linearized equation around a small pendulum
oscillation angle. However, the actual system is nonlinear, and the pendulum swings in a larger
angular region.
As an example, tuning of CPVA for six-cylinder engines is performed using the simulation
model. The resulting parameter R/b is 9.343, not 9 as suggested by Eq. 3.98. The simulation
result of the tuned CPVA is shown in Fig. 3.123. It should be noted that the pendulum swings
in the opposite direction of the engine torque fluctuations. In the top chart, the solid line is the
crank shaft speed without a CPVA; the dashed line is the crank shaft speed with a CPVA. The
speed fluctuation is greatly reduced by CPVAs.
3.8.5 Path of pendulum motion

The function in Eq. 3.91 defines the path of motion of the CM of pendulums. This section con-
siders several different path functions, and their effects on the CPVA performance.
The first path function is given in Eq. 3.103 and shown graphically in Fig. 3.124. It is an ellipti-
cal function. If a = b, then it is a circular path. Otherwise it is an oval path. By changing the
ratio of a/b, one can examine how it affects CPVA performance.
l = ab (a 2 cos 2 (φ ) + b 2 sin 2 (φ ))
− 12
(3.103)
Fig. 3.123 Simulation result of a tuned CPVA.
162
6293_Book.indb 162 1/11/13 3:54 PM

Fig. 3.124 Ellipse path

function.
The paths of epicycloids and cycloids have been under extensive investigations. The parameter-
ized functions of these two paths are shown in Eqs. 3.104 and 3.105, respectively.
l = y 2 + (x − R b −2r + b)2
 y 
φ = arctan  
 x − R − 2r + b 
4r (Rb + r )
b= (3.104)
2r + Rb
 R +r 
y = (Rb + r )sin(t ) + r sin  b t 
 r 
 R +r 
x = (Rb + r )cos(t ) + r cos  b t 
 r 
Here, Rb is the radius of the base cycle, r is the radius of the cycling cycle, and t is the free
parameter of the parameterized epicycloidal function.
r = y 2 + (2a + x )2
 y 
φ = arctan  
 2a + x  (3.105)
b = 4a
x = a (1 + cos(t ))
y = a (t + sin(t ))
Here, a is the radius of the cycling cycle, and t is the free parameter of the parameterized
cycloid function.
The last path function is given in Eq. 3.106, and is shown graphically in Fig. 3.125. It is an
even-term 4th order polynomial function. Let’s call it the ET4OP path. This path function
provides additional flexibilities to shape the pendulum motion. All other paths can be approxi-
mated using the ET4OP function.
l = b (1 + k1φ 2 + k2φ 4 ) (3.106)
163
6293_Book.indb 163 1/11/13 3:54 PM

Chapter 3
Fig. 3.125 ET4OP path function.
Figures 3.126–3.128 show the simulation comparison of different pendulum paths. Figure 3.126
is the simulation result of a circular motion pendulum. Figure 3.127 is the simulation result
of an epicycloid motion pendulum. Figure 3.128 is the simulation result of a cycloid motion
pendulum. In these figures, the first chart shows the crank shaft speed fluctuation with (broken
line) and without (solid line) CPVA, the second chart shows the engine torque, and the third
chart shows the angular displacement of the pendulum mass. It can be seen that the cycloid
path provides the best vibration reduction performance.
Fig. 3.126 Simulation result of a CPVA with a circular pendulum path.
164
6293_Book.indb 164 1/11/13 3:54 PM

Fig. 3.127 Simulation result of a CPVA with an epicycloid pendulum path.
Fig. 3.128 Simulation result of a CPVA with a cycloid pendulum path.
165
6293_Book.indb 165 1/11/13 3:55 PM

Chapter 3
In the previous equations, the variable b is the radius of curvature of the path at the point
of f = 0. The value of b/R determines the order of the CPVA. The performance sensitivity to
the changes of the value of b/R is another important characteristic to be evaluated for each
pendulum path. Lower sensitivity represents higher manufacturability. Figure 3.129 compares
the sensitivity of the circular, epicycloids, and cycloid path. In the figure, the horizontal axis is
the percentage change of b/R, the vertical axis is the percentage of crank shaft speed fluctua-
tion reduction. For circular path and epicycloidal path, the curves become vertical at the right
end. The vertical lines indicate that the pendulum becomes unstable. It can be seen from the
figure that the cycloid path is most insensitive to the changes of the value b/R.
Fig. 3.129 Comparison of performance sensitivity to changes of value of b/R.
166
6293_Book.indb 166 1/11/13 3:55 PM

Chapter 4
Metal Pushing V-Belt
Continuously Variable
Transmissions
167
6293_Book.indb 167 1/11/13 3:55 PM

Chapter 4
The Continuously Variable Transmission (CVT) allows more precise control of engine
operation points than conventional stepped-ratio transmissions; it therefore provides
potential for further improvements in vehicle fuel economy and drivability. Two basic
types of CVTs have been extensively developed: one is the metal pushing V-belt CVT
(V-CVT), another is the toroidal traction drive CVT (T_CVT). This chapter focuses
on ratio control technologies of V-CVTs. First, we will present some basic mechanics
of metal pushing V-belt CVTs. Then, the required pulley thrust force for achieving
the desired torque capacity and ratio position of V-CVTs will be discussed, followed
by a study of the hydraulic ratio control systems required to achieve such thrust force
controls. Two basic types of V-CVT ratio control, hydromechanical feedback control
and electronic feedback control, will be introduced. Finally, feed-forward and feedback
controls and their application to V-CVT ratio control will be presented.
4.1 Mechanics of metal pushing V-belt

continuously variable transmissions
The cross-section view of a metal pushing V-belt continuously variable transmission is shown
in Fig. 4.1. Besides the common components found in conventional step-gear automatic
transmissions, the main component in this cross section is the metal pushing V-belt pulley
system. A three-dimensional view of a metal pushing V-belt pulley system is shown in
Fig. 4.2.
Fig. 4.1 Cross-section view of a metal pushing V-belt CVT.
168
6293_Book.indb 168 1/11/13 3:55 PM

Metal Pushing V-Belt Continuously Variable Transmissions
Three major components make up a metal pushing V-belt pulley system: drive pulley
(primary pulley), driven pulley (secondary pulley), and V-belt. Each pulley is made of two
cone-shaped disks facing each other to form a V-shaped groove between them. One of the
two cone-disks can move axially to change the width of the V-groove. In Fig. 4.1, the left
cone of the drive pulley is axially movable, and the right cone of the driven pulley is axially
moveable. The V-belt rides on the V-grooves of the drive and driven pulleys. The riding
radiuses on the drive pulley and driven pulley are determined by the width of the V-grooves,
and therefore are determined by the axial position of the two axially movable cone-disks.
Assuming that the riding radius on the drive pulley is r p, and riding radius on the driven
pulley is rs, the speed ratio of the CVT, SRcvt, can be expressed as shown in Eq. 4.1. The details
of a metal pushing V-belt assembly are shown in Fig. 4.3. A V-belt assembly consists of the
multi-layer steel bands and the V-shaped metal pieces. Because the length of a V-belt is a
constant, when the riding radius of one pulley becomes smaller, the riding radius of the other
pulley becomes larger. The axially movable cones of the drive pulley and driven pulley are
pushed toward the fixed cones by the hydraulic cylinders. The thrust forces from those two
hydraulic cylinders control the position of the two movable cones, and therefore control the
speed ratio of the metal pushing V-belt pulley system. The torque capacity of the V-belt pulley
system is also established by those two thrust forces. An electrohydraulic control system is
used to control the thrust forces. In the next section, we will discuss the electrohydraulic
control systems of metal pushing V-belt pulley systems.
rs
SRcvt = (4.1)
rp
Fig. 4.2 Metal pushing V-belt pulley

system of a CVT.
Fig. 4.3 V-belt assembly.
169
6293_Book.indb 169 1/11/13 3:55 PM

Chapter 4
4.2 Controls of metal pushing V-belt

continuously variable transmissions
The torque capacity and speed ratio of a metal pushing V-belt CVT are controlled by the two
thrust forces Fp and Fs, which are exerted on the primary and secondary pulley, respectively,
as shown in Fig. 4.4. The required magnitude for each thrust force, Fp and Fs, to achieve the
desired torque capacity and speed ratio is a summation of two components: the torque-
capacity thrust force and the ratio thrust force [4-1].
The torque-capacity thrust force component for primary and secondary pulley thrust force
Fp and Fs are equal, and can be calculated from Eq. 4.2.
Tin cos a
Ftc = (4.2)
2mRp
Here, Tin is the transmittable primary pulley torque, a is the half-wedge angle of the pulley
V-groove, Rp is the pitch radius (riding radius) of the primary pulley, and m is the coefficient
of friction between the belt metal pieces and the pulley. The torque capacity thrust force is
relatively well defined and, therefore, can be determined mathematically with a satisfactory
precision.
To achieve a desired ratio position, in addition to the torque capacity thrust force, a certain
amount of ratio thrust force must be added to the primary pulley or secondary pulley based
on the desired speed ratio. Which pulley should have this additional ratio thrust force, and
the magnitude of it, are determined by many factors. The major factor is the kp/ks force ratio,
which is the required ratio of Fp/Fs (see Fig. 4.4) for maintaining the desired speed ratio, SRcvt.
The kp/ks force ratio for maintaining a desired speed ratio is a function of the speed ratio and
torque factor. Speed ratio is defined in Eq. 4.1. Torque factor is defined as the ratio of pulley
input torque to the torque capacity of the pulley. The kp/ks ratio is usually obtained from
hardware tests. Figure 4.5 is a graphical representation of atypical kp/ks ratio as a function of
speed ratio and torque factor. A certain amount of uncertainty of kp/ks exists in a real world’s
V-belt pulley system.
The ratio thrust force requested by the kp/ks force ratio is a steady-state thrust force,
meaning it is the thrust force to maintain the pulley at a desired ratio position. The dynamic
Fig. 4.4 Primary and secondary thrust forces

to control torque capacity and speed ratio.
170
6293_Book.indb 170 1/11/13 3:55 PM

ratio thrust forces, such as the inertial force and resistance force to be overcome during
ratio changing, are also significant parts of the ratio thrust force, and need to be counted.
Therefore, unlike the torque capacity thrust force, the ratio thrust force is difficult to deter-
mine mathematically with a satisfactory precision. This implies that the required hydraulic
pressure for generating the ratio thrust force is also difficult to determine mathematically
within a satisfactory precision.
From the discussions, we can conclude that it is desirable to design a hydraulic ratio control
system such that only the torque capacity thrust force (pressure) needs to be determined
mathematically. The other control elements such as ratio thrust force (pressure) and hydrau-
lic pump pressure are determined by some kind of automatic feedback controls, so those
control elements can be determined by the demands of the physical system itself directly in
real time.
The hydraulic ratio control system designed to achieve the desired features is shown in Fig. 4.6.
Ports A and B of the ratio valve are fed with torque capacity pressure. The torque capacity
pressure is controlled by the torque capacity pressure control valve and is commanded
directly by a microprocessor controller based on Eq. 4.2. The output pressure from port C
of the ratio valve is fed to the primary pulley cylinder, and the output pressure from port E
of the ratio valve is fed to the secondary pulley cylinder. Port D of the ratio valve is fed by
the pump pressure. The speed ratio is commanded by the ratio position input to the upper
end of the pulley position feedback lever. The lower end of the lever moves axially with the
movable cone-disk of the primary pulley. The middle point of the lever is connected to the
spool of the ratio valve. In this way, a hydromechanical position-feedback servo system is
formed. The control objective of the servo system is to let the axial position of the movable
cone-disk follow the ratio position input of the upper end of the lever. With this servo
control system in place, the cylinder to be fed by the torque capacity pressure only, and the
magnitude of the ratio thrust pressure adding to the other cylinder in addition to the torque
capacity pressure, are determined by the hydromechanical feedback loop automatically. The
Fig. 4.5 Typical kp/ks as function of speed ratio and torque factor.
171
6293_Book.indb 171 1/11/13 3:55 PM

Chapter 4
pump pressure control valve, together with the selection valve, maintains the pump pressure
just above the higher pressure amount of the primary and secondary pulley pressures by a
predetermined amount. This predetermined amount of pressure is provided by the spring in
the pump pressure control valve. The pump pressure control valve essentially is a differential
pressure control valve with a pressure balance equation of Ppump = max(Pprim,Psecond) + Pspring.
Here, Ppump is the pump pressure, Pprim is the primary pulley pressure, Psecond is the second-
ary pulley pressure, and Pspring is the pressure generated by the spring in the pump pressure
control valve. It is clear that the areas of the primary and the secondary pulley cylinders have
to be equal for proper operation of the system. The pump protection valve sets the upper limit
of pump pressure. Notice that due to the high hydraulic pressure requirement, the pump is
usually a fixed-displacement gear pump or equivalent.
With the described hydraulic ratio control system, the only control variables that need to be
commanded by the microprocessor controller are the torque capacity pressure signal and
ratio position input. The pump pressure and the ratio thrust pressure are determined by the
demands of the physical condition of the pulley system itself through the hydromechanical
ratio feedback control and the differential pressure control mechanism. Complicated calcula-
tions of the required ratio thrust force (pressure) are not necessary. Therefore, the kp/ks force
ratio is not used by the microprocessor control algorithms.
The position feedback lever shown in Fig. 4.6 can be eliminated, and the hydromechanical
feedback system can be replaced with an electronic feedback system. The hydraulic schematic
of such a variation of ratio control system is shown in Fig. 4.7.
Hydraulic ratio control systems can be analyzed using computer simulations. An AMESim
simulation model for this purpose is shown in Fig. 4.8. The upper part of the model is the
mechanical part of the V-CVT system, which includes the pulleys and belt, the input and
output shaft inertias, the primary and secondary cylinders, the piston mass, the character-
istics of kp/ks force ratio, and the centrifugal pressure. The lower part of the model is the
hydraulic ratio control system, which includes the ratio valve with hydromechanical pulley
Fig. 4.6 Hydraulic ratio control system with mechanical ratio feedback.
172
6293_Book.indb 172 1/11/13 3:55 PM

Fig. 4.7 Hydraulic ratio control system with electronic ratio feedback.
position feedback, the torque capacity pressure control valve, the pump pressure control
valve, the selection valve, the pump protection valve, the disturbances caused by the flow
demands from other hydraulic actuators, and the hydraulic pump. The pump is driven by the
input shaft with a speed fluctuation model.
Figure 4.9 is the simulation result of the ratio changing from under-drive to overdrive. The
actual pulley position follows the commanded pulley position well (see the second chart). The
tracking error of the pulley position is inherent to hydromechanical feedback systems, and
can be compensated with a modified pulley position command. The torque capacity pressure
(see the first chart) is commanded based on Eq. 4.2. The primary and secondary pulley pres-
sures are results of the hydromechanical ratio position feedback control. The pump pressure
is the result of the control of the pump pressure control valve. The primary and secondary
pressures provide just the right amount of thrust force to achieve the desired torque capacity
and ratio position. The pump pressure remains just higher than the higher pulley pressure by
a predetermined amount.
To demonstrate that the control system automatically adapts to the changes in kp/ks char-
acteristics, Fig. 4.10 shows the simulation results with changed kp/ks characteristics. The
pulley position follows command as well as the baseline simulation shown in Fig. 4.9. The
primary and secondary pulley pressure, and the pump pressure, automatically adapt to the
changed kp/ks.
Figure 4.11 is the simulation result of the ratio changing from under-drive to overdrive and
then returning to under-drive. The pulley position follows command well in both directions.
4.3 Comparison with other ratio control systems

This section compares the ratio control system described in the previous section with other
ratio control systems.
173
6293_Book.indb 173 1/11/13 3:55 PM

Chapter 4
Fig. 4.8 AMESim simulation model of proposed ratio control system.
Fig. 4.9 Simulation results of ratio change

from under-drive to overdrive.
174
6293_Book.indb 174 1/11/13 3:55 PM

Fig. 4.10 Simulation results with

different kp/ks characteristics.
Fig. 4.11 Simulation results of ratio

change from under-drive to overdrive,
and then from overdrive to under-
drive.
175
6293_Book.indb 175 1/11/13 3:55 PM

Chapter 4
First, let’s compare the system with a V-CVT ratio control system called “low-high-pressure
ratio control system.” The schematic of the low-high-pressure ratio control system is shown
in Fig. 4.12. In this system, a high pressure is fed to port D of the ratio valve, and a low
pressure is fed to ports A and B of the ratio valve. The high pressure and low pressure are
controlled by the torque pressure signal through the high-pressure control valve and low-
pressure control valve, respectively. There is a fixed ratio between high and low pressures,
which is determined by the gain of high- and low-pressure control valves. The controller
commands a torque pressure signal, which in turn produces the low and high pressures. The
low pressure is the torque capacity pressure that has to generate torque capacity thrust force,
as defined in Eq. 4.2. The high pressure is the pump pressure, which has to be high enough
to allow the ratio valve to provide the required ratio thrust force. It can be seen that the dif-
ference between this ratio control system and the system described in the previous section
is in how the pump pressure is controlled. In the previous system, the pump pressure is
controlled by the hydraulic feedback control system to the level just above the required pulley
pressure. In this low-high-pressure ratio control system, the pump pressure is commanded by
a microprocessor controller. Therefore, the calculations of both the required torque capacity
thrust force and ratio thrust force (including the kp/ks force ratio, the inertia force, etc.) have
to be performed in the microprocessor controller. Due to the uncertainties of those variables,
a larger safety factor must be used. This creates a greater potential to over pressurize the
system. In addition, in this system the high and low pressures have a fixed ratio. However, the
required force ratio kp/ks is not fixed but is a function of CVT speed ratio and torque factor.
Because the ratio between high and low pressures has to be designed to satisfy the maximum
kp/ks, the resulting system will be over pressurized at other kp/ks values. The drawbacks of
over-pressurized CVT ratio control systems are: increase of hydraulic power losses, decrease
of pulley mechanical efficiency, and reduction of durability of the belt.
The “master-slave ratio control system” is another alternative V-CVT ratio control system.
In the master-slave system, the area of primary pulley cylinder is two times the area of the
secondary pulley cylinder. The hydraulic schematic of the system is shown in Fig. 4.13. The
secondary cylinder is fed with the pump pressure, which is controlled by the pump pressure
control valve. The primary cylinder pressure is controlled by the ratio valve through a
Fig. 4.12 Low-high-pressure ratio control system.
176
6293_Book.indb 176 1/11/13 3:55 PM

Fig. 4.13 Master-slave system with hydromechanical pulley position feedback.
hydromechanical feedback mechanism. In this control system, the pump pressure, based on
the ratio position, sometimes determines the torque capacity thrust force, and sometimes
determines the ratio thrust force. Therefore, the determination of the pump pressure
command requires accurate knowledge of the required kp/ks ratio and dynamic forces; other-
wise, a large safety margin has to be used, which will lead to over pressurizing the system.
4.4 Feed-forward/feedback control and its

application to V-CVT ratio control
4.4.1 Introduction
Due to the existence of plant uncertainties and the time delay caused by microprocessor
controllers, the achievable bandwidth of a closed-loop control system is greatly limited. In
many cases in automatic transmission controls, the achievable bandwidth of closed-loop
controls, due to those reasons, is not wide enough to provide fast enough responses. To
improve the system response performance and also to improve robust stability, feed-forward
controls are used in conjunction with feedback controls. The applications of such feed-
forward and feedback control systems in automatic transmissions include torque converter
clutch slip speed controls, clutch-to-clutch gear shift controls, CVT ratio controls, etc. This
section introduces some fundamentals of feed-forward and feedback controls first, and then
discusses the applications in CVT ratio controls [4-2, 4-3, 4-4].
4.4.2 Limitations of feedback controls

Figure 4.14 shows a typical feedback control system. Here, C is the transfer function of
the feedback controller, and P is the transfer function of the plant, which is described by a
nominal plant Po and plant uncertainties DP, as shown in Eq. 4.3. Note that DP does not have
177
6293_Book.indb 177 1/11/13 3:55 PM

Chapter 4
Fig. 4.14 Block diagram of a

feedback control system.
any poles in the right-half complex plane. Also here, r is the reference signal, w is the distur-
bance signal, y is the output signal, and e is the error signal.
P = Po + DP (4.3)
The design requirement for a feedback control system includes two major aspects: control
performance requirement and robust stability requirement. To achieve a good control per-
formance, that is to achieve a small error signal e, the sensitivity transfer function S shown in
Eq. 4.4 shall be made as small as possible.
e e 1
S= = = (4.4)
r w 1 + PoC
To achieve a good, robust stability, the condition shown in the inequality of Eq. 4.5 has to
be satisfied. (The inequality can be proven using the Nyquist curve shown in Fig. 4.15.) This
means that the complementary sensitivity transfer function T shown in Eq. 4.6 shall be made
small enough to obtain sufficient, robust stability.
ΔP 1 + PoC
< (4.5)
Po PoC
PoC
T= (4.6)
1 + PoC
But,
T+S=1 (4.7)
Fig. 4.15 The Nyquist curve

to prove inequality Eq. 4.5.
178
6293_Book.indb 178 1/11/13 3:55 PM

This means that in the frequency range where |T(jw)| is made small (good, robust stability),
|S(jw)| will be large (larger error signal e and poor control performance), and vice versa. The
previous discussions conclude that the plant uncertainty imposes a great limitation on the
achievable control performance.
Microprocessor controllers introduce an additional time delay in the feedback control loops.
Considering this time delay, the block diagram of the feedback control system becomes the
one shown in Fig. 4.16.
The achievable bandwidth w gc of the feedback control system with a time delay t is given in
inequality Eq. 4.8 [4-3, 4-4].
π 1
ωgc < (4.8)
2τ
From the discussions, it can be seen that the existence of plant uncertainties and the time
delay caused by the microprocessor controller impose a great limitation to the achievable
control performance. In many cases in automatic transmission controls, this achievable
control performance is not fast enough to provide the desired control quality.
4.4.3 Feedback and feed-forward control systems

To improve the response to the reference input signal while maintaining sufficient stability
margins, a feed-forward control F can be added to the feedback controls, as shown in Fig.
4.17. The design of the feed-forward control F will be discussed in the following sections.
Characteristics of feed-forward controls include:
• Can be used to improve response to the reference signal

• Can be used to reduce the effect of measurable disturbance
• No risk of instability, because it is an open-loop control
• Design of feed-forward control is simple
• Should be used with feedback controls
Fig. 4.16 Block diagram of a feedback

control system with a time delay.
Fig. 4.17 Block diagram of a feedback/

feed-forward control system.
179
6293_Book.indb 179 1/11/13 3:55 PM

Chapter 4
Table 4.1 Properties of Feed-forward and Feedback Controls

Feed-forward Control Feedback Control
Open loop Closed loop
Act on reference signal Act on error signal
Not robust to model error Robust to model error
No risk of instability Risk of instability
Improved response to reference signal, or Compensate for plant variations and steady-
reduced effect to measurable disturbances state error
Feed-forward controls are open-loop controls, and cannot compensate for plant uncertain-
ties, steady state error, and initial conditions; therefore, they have to be used with feedback
controls. Table 4.1 shows some properties of feed-forward and feedback controls.
4.4.4 Design of feed-forward controls

One straightforward design of a feed-forward controller F(s) is as simple as shown in Eq. 4.9.
F (s) = Pm−1 (s) (4.9)
Here, Pm(s) is the nominal plant model. This approach has a few problems. First, if the plant
model Pm(s) has unstable zeros, then the feed-forward control F(s) will be unstable. Second,
if Pm (s) is strictly proper (most plant models are strictly proper), then F(s) will be improper
and requires differentiation. Third, if the plant is highly nonlinear, sometimes it is difficult to
obtain its inverse.
To improve feed-forward controls, a virtual feedback control based feed-forward control
system, as shown in Fig. 4.18, is proposed. Here, Cm(s) is the feedback controller of the
virtual feedback control system S(Pm(s), Cm(s)). The output signal of Cm(s) is fed to the actual
plant input as the feed-forward control signal. Because there are no plant uncertainties or
time delay in the virtual feedback control system S(Pm(s), Cm(s)), the bandwidth of S(Pm(s),
Cm(s)) can be designed wider than the real feedback control system S(P(s), C(s)). The virtual
feedback system S(Pm(s), Cm(s)) is designed to have the desired response to the reference
signal r, such that the real plant responds to the reference signal r in parallel with the same
output signal. The real feedback control system S(P(s), C(s)) is designed to have enough stabil-
ity margins, good control performance in the small error region, and satisfactory steady-state
error requirements. The virtual feedback control system S(Pm(s), Cm(s)) also works as the
Smith compensator to compensate for the time delay.
When there is a one-to-one corresponding relationship between the steady state of the plant
input and its outputs, the virtual feedback control based feed-forward controller can be
simplified as shown in Fig. 4.19.
In Fig. 4.19, the feed-forward controller is a first-order system. The gain k determines the
time constant of the first-order system. The feed-forward control signal v has two compo-
nents. One is the function of the output signal of the first-order system, and is defined by the
function F1; the other is the function of the derivative of the output signal, and is defined by
180
6293_Book.indb 180 1/11/13 3:55 PM

Fig. 4.18 Virtual feedback based

feed-forward and feedback control
system.
Fig. 4.19 A variant of the virtual

feedback based feed-forward and
feedback control system.
the function F2. F1 can be obtained from inversing the steady-state gain of the plant. F2 is
designed to generate a counter-plant damping force, which is a function of the derivative of
the plant output.
4.4.5 Application: direct pulley pressure control of V-CVT

The direct pulley pressure control of metal pushing V-belt CVTs is a hydraulic control system
that has two pressure control valves directly controlling the primary and secondary pulley
pressures. There is no hydromechanical ratio position feedback. The primary and secondary
pulley pressures are commanded by the microprocessor controller, and are directly based on
the desired speed ratio, the current speed ratio, and required torque capacity. The hydraulic
schematic of this type of control system is shown in Fig. 4.20. The control signal pressures
for primary pulley pressure control valve, secondary pulley pressure control valve, and line
pressure control valve are from three separate pressure control solenoids (not shown in the
figure). These three pressure control solenoids are directly commanded by a microprocessor
controller.
In a V-CVT with direct pulley pressure controls, there is no hydromechanical ratio position
feedback control. Speed ratio of the V-CVT is controlled by a microprocessor controller,
which directly commands primary and secondary pressures. An electronic ratio feedback
control algorithm in the microprocessor generates this command. A block diagram of the
V-CVT direct pressure control algorithm using the proposed feed-forward and feedback
control system is shown in Fig. 4.21. In the figure, the block labeled CVT is the plant model.
The input to the plant model is the kp/ks ratio, which is the thrust force ratio of the primary
pulley and secondary pulley. The plant model includes the CVT pulley-belt, hydraulic
control system, and the algorithm for calculating the primary and secondary pulley pressure
command based on the input signal kp/ks and the CVT input torque (the algorithm will
be described later). The output from the plant model is the CVT speed ratio. The rest of the
block diagram is the feed-forward/feedback controller for V-CVT ratio control. The feedback
controller used here is a PI controller with a scheduled proportional gain and a selectable
integral gain. The integral gain is selected based on the error signal of the feed-forward
control system. The saturation block in the integral control loop is for anti-windup purposes.
181
6293_Book.indb 181 1/11/13 3:55 PM

Chapter 4
Fig. 4.20 Hydraulic schematic of direct pulley pressure control system.
The feed-forward function F1 is the steady-state mapping function from speed ratio to kp/ks
ratio at 70% torque capacity ratio. The feed-forward function F2 is the plant’s damping coef-
ficient, including hydraulic resistances and friction. The output of the feed-forward/feedback
controller is kp/ks. The primary and secondary pulley pressure commands are calculated
from the kp/ks ratio, as shown in Eqs. 4.10–4.14.
Pprim = (Max(Fclamp,Fclamp · kpks)Fcnet_prim)/Aprim (4.10)
Psecd = (Max(Fclamp,Fclamp / kpks) – Fcent_secd)/Asecd (4.11)
Tin cos a
Fclamp = ⋅ (1 + c ) (4.12)
2 mRp
Fcent _ prim = kprim w prim

2
(4.13)
Fcent _ secd = ksecd w secd

2
(4.14)
Here, Pprim is the primary pulley pressure command, Psecd is the secondary pulley pressure
command, and Fclamp is the pulley thrust force for obtaining the desired torque capacity. Tin
is the input torque, c is the safety factor that usually takes a value of 0.3, a is the half-wedge
angle of the pulley groove, Rp is the pitch radius of the primary pulley, and m is the coefficient
of friction between belt blocks and pulleys. Fcent_prim is the centrifugal force of the primary
pulley, kprim is the centrifugal force coefficient of the primary pulley, and w prim is the angular
velocity of the primary pulley. Fcent_secd is the centrifugal force of the secondary pulley, ksecd
is the centrifugal force coefficient of the secondary pulley, w secd is the angular velocity of the
secondary pulley, Aprim is the area of the primary cylinder, and Asecd is the area of the second-
ary cylinder.
The comparison between the systems with and without feed-forward controls is given in Figs.
4.22 and 4.23. Figure 4.22 is the simulation result of feedback controls only, and Fig. 4.23 is
182
6293_Book.indb 182 1/11/13 3:55 PM

the simulation result of feedback and feed-forward controls. The feedback control is designed
to have enough of a stability margin to provide robust stability against system variations.
Therefore, the feedback control itself cannot provide a fast enough time response. Adding
the feed-forward controls greatly improves time response performance while maintaining
robust stability.
Fig. 4.21 Block diagram of CVT ratio direct pressure control using proposed feed-forward and feedback control system.
Fig. 4.22 Simulation result of time

response to a ratio command of feedback
only system.
183
6293_Book.indb 183 1/11/13 3:55 PM

Chapter 4
Fig. 4.23 Simulation result of time

response to a ratio command of feedback
and feed-forward system.
184
6293_Book.indb 184 1/11/13 3:55 PM

Chapter 5
Dynamics and Controls of
Dual-Clutch Transmissions
185
6293_Book.indb 185 1/11/13 3:55 PM

Chapter 5
The Dual-Clutch Transmission (DCT) is a different type of automatic transmission.

Although DCTs cannot perform all types of power-on gear shifts, as plenary gear auto-
matic transmissions can ( DCTs cannot perform odd-gear-to-odd-gear power-on shifts,
or even-gear-to-even-gear power-on shifts), due to their higher efficiency, DCTs are
becoming increasingly attractive. Manufacturers or regions that are more familiar with
manual transmission technologies also tend to choose DCT as the next step for auto-
matic transmission technology.
5.1 Construction of dual-clutch transmissions

Figure 5.1 is a stick diagram of a dual-clutch transmission. The input shaft is split into two
power paths. One power path can be connected to the odd gear shaft if the frictional clutch
C1 is applied. Another power path can be connected to the even gear shaft if the frictional
clutch C2 is applied. The gear ratio between the odd gear shaft and the output shaft can be
selected by controlling the position of the 1-3 synchronizer and 5-7 synchronizer (synchroniz-
ers will be described later). For example, if the 1-3 synchronizer is at the 1st gear position, and
the 5-7 synchronizer is (and must be) at the neutral position, then the gear ratio between the
odd gear shaft and the output shaft is 1st gear. The gear ratio between the even gear shaft and
the output shaft can be selected in the same manner. To achieve a desired input-output gear
ratio, the corresponding synchronizer and clutch have to be applied. For example, 1st gear is
achieved by engaging the 1st gear synchronizer and applying clutch C1 (C2 has to be released);
2nd gear is achieved by engaging the 2nd gear synchronizer and applying clutch C2 (C1 has to
be released). Other gears can be achieved in the same manner. Shifting between an odd gear
and an even gear is accomplished by controlling the torque capacity of frictional clutches C1
and C2. The mechanics of such gear-shift controls is the same as that discussed in the gear-
shift mechanics of planetary gear automatic transmissions. Therefore, power-on shift could be
achieved between an odd gear and an even gear. Shifting between two odd gears (or between
two even gears) has to be performed with engine power off.
The architecture of the DCT shown in Fig. 5.1 is very straightforward; however, it is not attrac-
tive from a cost and packaging perspective due to its content-heavy design. A more-compact
architecture is shown in Fig. 5.2. In this architecture, the frictional clutches C1 and C2 are
arranged concentrically. Some of the gears are shared by multiple gear ratios.
Fig. 5.1 Stick diagram of a

dual-clutch transmission.
186
6293_Book.indb 186 1/11/13 3:55 PM

Dynamics and Controls of Dual-Clutch Transmissions
Fig. 5.2 Stick diagram of a DCT

with a more-compact architecture.
5.2 Synchronizer and its control

As described in the previous section, synchronizers are used to preselect gears. The construc-
tion of synchronizers used in DCTs is the same as the one used in manual transmissions; it is
shown in Fig. 5.3. The left gear assembly and right gear assembly are rotationally free on the
shaft if the hub-sleeve assembly is at the central position (neutral position). The hub is con-
nected to the shaft via a spline; therefore, it rotates with the shaft and is translatable in the axle
direction on the shaft. The sleeve is connected to the hub via a spline in the same manner. To
couple the left gear assembly with the shaft, the sleeve is pushed leftward. Due to the force
of the detent ball, the hub will move leftward with the sleeve. As the hub is moving leftward,
the outer corn and inner corn will be contacted. The friction torque between the outer corn
and inner corn will synchronize the speed of the left gear assembly and shaft. At this point,
although the hob is stopped by the outer corn, the sleeve can still continue to move leftward.
As the speed of the left gear assembly and the shaft are synchronized, the inner dog teeth will
be mashed into the outer dog teeth, and the left gear assembly is coupled with the shaft. Figure
5.4 shows the synchronizer at the left gear engaged position. To couple the right gear assembly
with the shaft, the sleeve is pushed rightward.
In DCTs, the position of synchronizers is controlled by linear actuators. There are two basic
types of such linear actuators: hydraulic actuators and electromechanical actuators.
Fig. 5.3 Construction of synchronizers.
187
6293_Book.indb 187 1/11/13 3:55 PM

Chapter 5
Fig. 5.4 Synchronizer at left gear

selected position.
Fig. 5.5 Synchronizer controlled by

a hydraulic actuator.
Figure 5.5 shows a synchronizer controlled by a linear hydraulic actuator. The synchronizer
control fork rides on the guide rail. As the hydraulic actuator pushes the synchronizer control
fork left or rightward, the sleeve and synchronizer move along with it.
Two types of hydraulic linear actuators for synchronizer controls have been developed. One
has two stable positions, as shown in Fig. 5.6. The other has three stable positions, as shown
in Fig. 5.7. For the three-position actuator, when p1 is fed with a high pressure and p2 is
exhausted, the actuator moves to the right end; when p2 is fed with a high pressure and p1 is
exhausted, the actuator moves to the left end; when both p1 and p2 are fed with the same high
pressure, the actuator moves to the central position. For the two-position actuator, because
there no stable central position, a linear position sensor and closed-loop position control are
necessary to achieve three position controls.
Fig. 5.6 Hydraulic linear actuator with

two stable positions.
188
6293_Book.indb 188 1/11/13 3:55 PM

Fig. 5.7 Hydraulic linear actuator with

three stable positions.
Fig. 5.8 Synchronizer controlled by an

electromechanical actuator.
Another type of linear actuator for synchronizer controls is the electromechanical actuation
system. A typical construction of such systems is shown in Fig. 5.8. An electric motor drives
a barrel cam through speed-reduction gears. The barrel cam converts rotary motion to linear
motion of the synchronizer control fork. A position sensor is necessary to control an electro-
mechanical actuation system.
5.3 Dual-clutch module

A dual-clutch module consists of two frictional clutches C1, C2 and clutch actuation
systems. There are two types of dual-clutch modules: wet dual-clutch modules and dry dual-
clutch modules.
Figure 5.9 shows a wet dual-clutch module. As shown in Fig. 5.9, the apply force of wet
clutch C1 is determined by the hydraulic pressure exerted on piston 1. When clutch C1 is
applied, shaft 1 is connected to the engine input. Clutch C2 is controlled in the same manner
by piston 2, and connects shaft 2 to the engine input. Hydraulic pressure for controlling wet
dual clutches is usually controlled by a pressure regulation system, as described in previous
sections. Wet clutches are immersed in transmission fluid for cooling and cleaning, and there-
fore have better thermo capacity and are easier to control in comparison with dry clutches.
Due to the low friction coefficient, a wet clutch usually has multiple clutch plates. Adding to
the fact that the transmission fluid between clutch plates causes drag force in wet clutches, wet
clutches have more drag loss when in the released state than dry clutches.
189
6293_Book.indb 189 1/11/13 3:55 PM

Chapter 5
Fig. 5.9 Wet dual-clutch module.
Most dry dual-clutch modules adapted from the clutch design of manual transmissions. Figure
5.10 shows a typical design of dry dual-clutch modules. Due to the higher friction coefficient,
a dry clutch only needs one clutch plate; therefore, the drag loss of dry clutches is much lower
than wet clutches. In this particular design shown in Fig. 5.10, similar to manual transmission
clutch designs, diaphragm spring levers are used to apply the clutch. Clutch actuation forces
push on one end of the spring levers and cause deformations. The level of deformation of the
diaphragm spring lever determines the clutch apply force. Closed-loop position feedback
controls are usually used to control the position of clutch-apply bearings, and in turn to
control deformation of the spring lever and clutch-apply force.
Figure 5.11 shows a block diagram of closed-loop position feedback control of a spring-
lever-applied dry clutch. In the figure, Fd is the desired clutch apply force, Xc is the hydraulic
actuator position command (desired apply bearing position), Xm is the measured hydraulic
actuator position, and Q is the control signal for the flow control valve. Fd is converted to Xc
by the clutch-apple force to position the command conversion block. The closed-loop position
control algorithm controls the hydraulic flow control valve to ensure that Xm follows Xc with
the desired response time, tracking and steady-state error, and robust stability. A position
transducer is used to measure the position of the apply bearing.
There are some issues with such closed-loop, position-feedback-control-based, spring-lever-
applied, dry clutch control systems. First, the system is expensive due to the necessity of the
position transducer and position feedback control algorithm. Second, due to variations of
the relationship between clutch-apply force and apply bearing position caused by individual
build variations and wearing of clutch plates, even with a perfect position control in place, the
clutch-apply force still is subject to large variations.
To overcome the problems associated with diaphragm, spring-lever-applied, dry clutch control
systems, a direct force control system as shown in Fig. 5.12 is proposed. The system replaced
the traditional diaphragm spring lever with a stiffened diaphragm spring lever. The stiffened
lever enables direct force control of dry clutches. The same hydraulic pressure control systems
widely used in traditional automatic transmissions can be used to perform such direct force
control. Due to less variation between hydraulic pressure and clutch-apply force, the control is
more accurate. In addition, because there is less compliance between actuator and clutch plate,
190
6293_Book.indb 190 1/11/13 3:55 PM

Fig. 5.10 Dry dual-

clutch module.
Fig. 5.11 Closed-loop

control of apply bearing
position for diaphragm
spring-lever-applied dry
clutch system.
Fig. 5.12 Direct force

control for stiffened
diaphragm spring-lever-
applied dry clutch.
191
6293_Book.indb 191 1/11/13 3:55 PM

Chapter 5
Fig. 5.13 Rigid-lever-applied

dry dual-clutch module.
the control is more responsive. The internal model based force control block shown Fig. 5.12
converts the desired apply force Fd to a hydraulic pressure command p, with considerations of
dynamic characteristics between pressure command and clutch-apply force.
A rigid-lever-applied dry clutch control system shown in Fig. 5.13 is also proposed for achiev-
ing the same objectives as stiffened diaphragm spring levers. Some designs use the rigid lever
mechanism in wet dual-clutch controls.
5.4 Control algorithms for DCTs

The following major control algorithms are necessary for DCT controls.
1. Friction launch control algorithm
2. Gear shift schedule algorithm
3. Gear selection control algorithm (synchronizer control algorithm)
4. Gear shift control algorithm.
These control algorithms, other than number 3, are similar to those used in planetary gear
automatic transmissions, and have been discussed in the previous sections.
The gear selection control (synchronizer control) algorithm is unique to DCTs. Once the gear
shift schedule algorithm generates a new desired gear command, the gear selection control
algorithm controls the synchronizer actuators to preselect the desired gear. The synchronizer
actuator control consists of two phases. The first phase is the fast forwarding phase. In this
phase, the synchronizer is moved as fast as possible to make the inner corn contact the outer
corn (see Figs. 5.3–5.5). The second phase of control is the synchronization phase. In this
phase, an adequate force is applied to the synchronizer control fork to ensure a smooth syn-
chronization and meshing of the inner dog teeth and outer dog teeth.
192
6293_Book.indb 192 1/11/13 3:55 PM

5.5 Conclusion
Control of automatic transmissions—including the planetary gear automatic transmission
(AT), dual-clutch transmission (DCT), and continuously variable transmission (CVT)—is
executed by real-time control software embedded in microprocessor controllers. Very often
the questions are asked, “What is the adequate process for developing real-time control
software for automatic transmissions to achieve optimal control performance?” and “What
fields of expertise are necessary for achieving such an objective?” It is the authors’ belief that
development of real-time control software should start from understanding the fundamental
physics of the controlled system, understanding the dynamic as well as static characteristics
of the system, understanding the expectation of customers, and therefore understanding the
control objectives. From a thorough understanding of all those aspects, adequate software
architecture can be determined, simple and intelligent control algorithms can be designed, and
optimized control software can be developed. It is the authors’ hope that this book provides
the basic tools and knowledge for dynamic analysis and control system design (includ-
ing mechanical, electrical, and hydraulic control hardware as well as control software) of
automatic transmissions.
193
6293_Book.indb 193 1/11/13 3:55 PM

6293_Book.indb 194 1/11/13 3:55 PM

References
2-1 Benford, H. L., and M. B. Leising, “The lever analogy: a new tool in transmission
analysis,” SAE Paper No. 810102, SAE International, Warrendale, PA, 1972.
2-2 Mägi, Mart, “Classical Planetary Gear Train Problems Treated by Novel Powerful
Analytical Methods,” Proceedings of FISITA 2006, F2006P256.
2-3 Mathis, R., and Remond, Y., “A New Approach to Solving the Inverse Problem for
Compound Gear Trains,” ASME, J. Mech. Des., 121, pp. 98–106, 1999.
3-1 Merritt, Herbert Eugen, Hydraulic Control System, John Wiley & Sons, Inc., 1967.
3-2 Tsutsui, Hiroshi, Takayuki Hisano, Akitomo Suzuki, Makoto Hijikata, Masatoshi
Taguchi, and Koichi Kojima, “Electro-hydraulic control system for Aisin AW new
6-speed automatic transmission,” SAE Paper No. 2004-01-1638, SAE International,
Warrendale, PA, 2004.
3-3 Watechagit, Sarawoot, and Krishnaswamy Srinivasan, “Modeling and simulation of a
shift hydraulic system for a stepped automatic transmission,” SAE Paper No. 2003-01-
0314, SAE International, Warrendale, PA, 2003.
3-4 Bai, Shushan, Robert L. Moses, Todd Schanz, and Michael J. Gorman, “Development
of a new clutch-to-clutch shift control technology,” SAE Paper No. 2002-01-1252, SAE
International, Warrendale, PA, 2002.
3-5 Leigh, James Ronald, Control Theory, Institution of Electrical Engineers, 2004.
3-6 Bellman, Richard Ernest, Dynamic Programming, Courier Dover Publications, 2003.
3-7 Levine, William S., The Control Handbook, CRC Press, IEEE Press, 1996.
3-8 Bai, Shushan, Joel Maguire, Kim Daekyun, and Huei Peng, “Dynamic Programming
Based Motor Vehicle Shift Map Generation,” Proceedings of FISITA 2006, F2006P029.
3-9 Kim, D., H. Peng, S. Bai, and J. M. Maguire, “Control of Integrated Powertrain With
Electronic Throttle and Automatic Transmission,” IEEE Transactions on Control
System Technology, May 2007, Volume 15, Issue 3, P474–482.
3-10 Tat Pham, Trung, and Guanrong Chen, Introduction to Fuzzy Sets, Fuzzy Logic, and
Fuzzy Control Systems, CRC Press, 2000.
3-11 De Silva, Clarence W., Intelligent Control: Fuzzy Logic Applications, CRC Press, 1995.
195
6293_Book.indb 195 1/11/13 3:55 PM

References
3-12 Bastian, Andreas, “The potential of fuzzy logic in automatic transmission control,”
JSAE Paper No. 9438330, Oct. 1994.
3-13 Hirako, Atsushi, Akira Takayama, Yasuhiko Dote, and Kohki Hayashi, “The
modification of shift map using fuzzy logic,” JSAE Paper No. 912290, Oct. 1991.
3-14 Bai, Shushan, Daniel Brennan, Don Dusenberry, Xuefeng Tao, and Zhen Zhang‚
“Integrated Powertrain Control,” SAE Paper No. 2010-01-0368, SAE International,
3-15 Goetz, M., M. C. Levesley, and D. A. Crolla, “Integrated powertrain control of
gearshifts on twin clutch transmissions,” SAE Paper No. 2004-01-1637, SAE
3-16 Kuwahara, Seiji, Hideki Kubonoya, Hiroshi Mizuno, Masato Kaigawa, and Katsumi
Kono, “Toyota’s New Integrated Drive Power Control System,” SAE Paper No. 2007-01-
4-1 Lee, Heera, and Hyunsoo Kim, “Analysis of primary and secondary thrusts for a
metal-belt CVT—Part 1: New formula for speed ratio-torque-thrust relationship
considering band tension and block compression,” SAE Paper No. 2000-01-0841, SAE
4-2 Levine, William S., The Control Handbook, CRC Press, IEEE Press, 1996.
4-3 Astrom, Karl J., and Tore Hagglund, “PID Controllers: Theory, Design, and Tuning,”
International Society of Automation,1995.
4-4 Astrom, Karl J., and Bjorn Wittenmark, Computer-Controlled Systems: Theory and
Design (3rd Edition), Prentice Hall, 1996.
196
6293_Book.indb 196 1/11/13 3:55 PM

Bibliography
Chapter 2
Arnold, V. I., Mathematical Methods of Classical Mechanics, Springer, 1997.
Baran, Jeff, James Hendrickson, and Michael Solt, “General Motors New Hydra-Matic
RWD Six-Speed Automatic Transmission Family,” SAE Paper No. 2006-01-0846, SAE
Deur, Josko, Jahan Asgari, and Davor Hrovat, “Modeling of an automotive planetary gear set
based on Karnopp model for clutch friction,” IMECE2003-41693, pp. 903–910.
Haj-Fraj, Ali, and F’riedrich Pfeiffer, “Dynamic Modeling and Analysis of Automatic
Transmissions,” Proceedings of the 1999 EEWASME, International Conference on
Advanced Intelligent Mechatronics, September 19–23, 1999, Atlanta, USA.
Hedman, A., “Transmission Analysis—automatic derivation of relationships,” J. Mech. Des.,
December 1993, Volume 115, Issue 4, 1031.
Jandasek, V. J., “The design of a single-stage three-element torque converter,” SAE Paper No.
Kasuya, Satoru, Takao Taniguchi, Kazumasa Tsukamoto, Masahiro Hayabuchi, Masaaki
Nishida, Akitomo Suzuki, and Hiroshi Niki, “AISIN AW New High Torque Capacity Six-
Speed Automatic Transmission for FWD vehicles,” SAE Paper No. 2005-01-1020, SAE
Kelley, Oliver K., “The design of planetary gear trains,” SAE Paper No. 590059, SAE
Martin, G. H., Kinematics and Dynamics of Machines, McGraw-Hill, New York, NY, 1982.
Naunheimer, Harald, Bernd Bertsche, Joachim Ryborz, and Wolfgang Novak, Automotive
Transmissions, Springer, 1994.
Nozaki, Kazutoshi, Yuuji Kashihara, Nobuaki Takahashi, Akira Hoshino, and Atsushi Mori,
“Toyota‘s new five-speed automatic transmission A750E/A750F for RWD vehicles,” SAE
Paper No. 2003-01-0595, SAE International, Warrendale, PA, 2003.
197
6293_Book.indb 197 1/11/13 3:55 PM

Bibliography
Nozaki, Yoshinobu, Yoshikazu Tanaka, Hideo Tomomatsu, Hiroyuki Tsukamoto, and Futomi
Hanji, “Toyota‘s new six-speed automatic transmission A761E for RWD vehicles,” SAE
Oshidari, Toshikazu, Noboru Hattori, and Yoshirou Morimoto, “A new approach to finding
optimum planetary gear trains for automatic transmissions,” SAE Paper No. 930676, SAE
Ross, Craig S., “A method for selecting parallel-connected planetary gear train arrangements
for automotive automatic transmissions,” SAE Paper No. 911941, SAE International,
Runde, J. K., “Modeling and control of an automatic transmission,” SMME Thesis, Department
of Mechanical Engineering, MIT, Boston, Massachusetts, 1986.
Salgado, David R., and J. M. Del Castillo, “Selection and Design of Planetary Gear Trains
Based on Power Flow Maps,” Journal of Mechanical Design (Transactions of the ASME),
Vol. 127, no. 1, pp. 120–134, Jan. 2005.
Scherer, Heribert, “ZF 6-speed automatic transmission for passenger cars,” SAE Paper No.
2003-01-0596, SAE International, Warrendale, PA, 2003.
Seo, Jungmin, and Seung-Jong Yi, “Design of an automatic transmission system having an
arbitrary power flow using the automatic power flow generation algorithm,” Proceedings of
the Institution of Mechanical Engineers D, Journal of Automobile Engineering, Vol. 219, no.
D9, pp. 1085≠1097, Sept. 2005.
Tanaka, Michihiko, “Matrix methods in planetary gear train analyses,” SAE Paper No. 841190,
SAE International, Warrendale, PA, 1984.
Tian, Lei, and Lu Li-qiao, “Matrix system for the analysis of planetary transmissions,” Journal
of Mechanical Design, Transactions of the ASME, v 119, n 3, Sept. 1997, pp. 333–337.
Tsai, Lung-Wen, Mechanism Design: enumeration of kinematic structures according to function,
CRC Press, 2001.
Wilson, C. E., J. P. Sadler, and W. J. Michels, Kinematics and Dynamics of Machinery, Harper &
Row, New York, NY, 1983.
Chapter 3
Alsuwaiyan, A. H., and S.W. Shaw, “Performance and dynamic stability of general-path
centrifugal pendulum vibration absorbers,” Journal of Sound and Vibration 252, pp.
791–815, 2002.
Benman, H. H., “Tautochronic Bifilar Pendulum Torsion Absorbers for Reciprocating
Engines,” Journal of Sound and Vibration, 159(2), 251–277, 1992.
Cai, Y., and T. Hayashi, “The linear approximated equation of vibration of a pair of spur gears,”
Trans. ASME, J. Mech. Des., 1994, 116, 2, 558–564.
Cho, D., and J. K. Hedrick, “Automotive powertrain modeling for control,” Trans. ASME, J.
Dynamic System, 1989, 111, 568–576.
DeJonge, Bob, “Electronically controlled interactive shift control system for motor vehicles,”
SAE Paper No. 2001-01-2477, SAE International, Warrendale, PA, 2001.
198
6293_Book.indb 198 1/11/13 3:55 PM

Bibliography
Deur, J., J. Asgari, and D. Hrovat, “Modeling and analysis of automatic transmission
engagement dynamics—Nonlinear case including validation,” Trans. ASME, J. Dyn. Syst.
Meas. Control, vol. 128, pp. 251–262, Jun. 2006.
Deur, J., J. Asgari, D. Hrovat, and P. Kovač, “Modeling and analysis of automatic transmission
engagement dynamics—Linear case,” Trans. ASME., J. Dyn. Syst. Meas. Control, vol. 128,
pp. 263–277, Jun. 2006.
Deur, Josko, Josko Petric, Jahan Asgari, and Davor Hrovat, “Recent Advances in Control-
Oriented Modeling of Automotive Power Train Dynamics,” IEEE/ASME Transactions on
Mechatronics, Vol. 11, No. 5, October 2006, 513.
Erjavec, Jack, Automatic Transmissions, Thomson Delmar Learning, 2004.
Fujita, Kenjiro, and Toyoji Onishi, “Advanced shift control technology on newly-developed
automatic transaxle,” JSAE Review, Vol. 16, Issue 3, p. 310, July 1995.
Gibson, Alex, and Ilya Kolmanovsky, “Modeling and Analysis of Engine Torque Modulation
for Shift Quality Improvement,” SAE Paper No. 2006-01-1073, SAE International,
Higashimata, Akira, Kazutaka Adachi, Satoshi Segawa, Nobuo Kurogo, and Hironobu Waki,
“Development of a slip control system for a lock-up clutch,” SAE Paper No. 2004-01-1227,
SAE International, Warrendale, PA, 2004.
Inagawa, Tomokazu, Hideo Tomomatsu, Yoshikazu Tanaka, Kazuyuki Shiiba, and Hisanori
Shirai, “Shift control system development (NAVI·AI-SHIFT) for 5-speed automatic
transmissions using information from the vehicle’s navigation system,” SAE Paper No.
2002-01-1254, SAE International, Warrendale, PA, 2002.
Kawai, Masao, Hideki Aruga, Kunihiro Iwatsuki, Takashi Ota, and Takeo Hamada,
“Development of a shift control system for automatic transmissions using information
from a vehicle navigation system,” SAE Paper No. 1999-01-1095, SAE International,
Kondo, Masami, Yoshio Hasegawa, Yoji Takanami, Kenji Arai, Masaharu Tanaka, and
Masafumi Kinoshita, “Toyota AA80E 8-Speed Automatic Transmission with Novel
Powertrain Control Systems,” SAE Paper No. 2007-01-1311, Detroit, Michigan, April
16–19, 2007.
Mogalapalli, Srinivas N., Edward B. Magrab, and Lung-Wen Tsai, “A CAD system for the
optimization of gear ratios for automotive automatic transmissions,” SAE Paper No.
Nester, T. M., P. M. Schmitz, A. G. Haddow, and S. W. Shaw, “Experimental Observations
of Centrifugal Pendulum Vibration Absorbers,” Proceedings of the 10th International
Symposium on Transport Phenomena and Dynamics of Rotating Machinery, March 2004.
Newland, David Edward, “Nonlinear Vibration: A Comparative Study with Applications to
Centrifugal Pendulum Vibration Absorbers,” MIT Doctoral Dissertation, May 1963.
Sawamura, K., Y. Saito, S. Kuroda, J. Takahashi, and A. Katoh, “ Development of an Integrated
Power Train Control System with an Electronically Controlled Throttle,” JSAE Review,
Volume 17, Number 4, October 1996, pp. 446–446(1).
Seron, M., J. Braslavsky, and Graham C. Goodwin, Fundamental limitations in filtering and
control, Springer, 1997.
199
6293_Book.indb 199 1/11/13 3:55 PM

Bibliography
Shaw, S. W., V. Garg, and C. P. Chao, “Attenuation of Engine Torsional Vibrations Using Tuned
Pendulum Absorbers,” SAE Paper No. 971961, SAE International, Warrendale, PA, 1997.
Tokura, Takaaki, Tomohiro Asami, Yoshio Hasegawa, Toshio Sugimura, Katsumi Kono,
and Kenji Aoki, “Development of Smooth Up-Shift Control Technology for Automatic
Transmissions with Integrated Control of Engine and Automatic Transmission,” SAE Paper
No. 2007-01-1310, SAE International, Warrendale, PA, 2007.
Yano, Candace Arai, A Forward Dynamic Programming Approach for General Uncapacitated
Multi-stage Lot-sizing Problems, University of Michigan, College of Engineering,1986.
Chapter 4
Bents, David J., “Axial force and efficiency tests of fixed center variable speed belt drive,” SAE
Paper No. 810103, SAE International, Warrendale, PA, 1981.
Chapter 5
Matthes, Bernd, “Dual Clutch Transmissions—Lessons Learned and Future Potential,” SAE
Wheals, J. C., A. Turner, K. Ramsay, A. O’Neil, J. Bennett, and Haiping Fang, “Double Clutch
Transmission (DCT) Using Multiplexed Linear Actuation Technology and Dry Clutches
for High Efficiency and Low Cost,” SAE Paper No. 2007-01-1096, SAE International,
Wheals, J. C., J. McMicking, S. Shepherd, B. Bonnet, N. Jackson, and O. Hall, “Proven High
Efficiency Actuation and Clutch Technologies for eAMT and eDCT,” SAE Paper No. 2009-
01-0513, SAE International, Warrendale, PA, 2009.
200
6293_Book.indb 200 1/11/13 3:55 PM

Index
Acceleration estimators path of pendulum motion, 162–166

filter-observer based, 116–118 Simulink model of, 161
Kalman, 114–116 tuning of, 162
Algebraic analysis of planetary gear trains, 26–29 Clutch
AMESim models apply force of, 39
hydraulic clutch control system, 91 dual-clutch models, 57–62
hydraulic ratio control system, 172 synchronization of, 41
Applying mode, 122 torque capacity of, 25, 38–39
Artificial intelligence–based shift scheduling Clutch capacity, 46
system, 142–150 Clutch control
fuzzy controls, 145–148 electronic torque converter, 121–126
fuzzy logic operations, 144 off-going methods, 111
fuzzy logic rules for up-shift and down-shift see also Hydraulic clutch control systems
points, 148–150 Clutch fill detection, 114
fuzzy set, membership function, and truth Clutch fill phase, control of, 111
value, 143–144 Clutch-to-clutch gear-shift control
Automatic transmissions acceleration estimator, 114–118
basics of, 1–7 canceled shifts, 118–119
continuously variable transmissions (CVTs), 6 clutch fill detection, 114
six- and eight-speed planetary, 65–69 control of power-on down-shifts, 112
step gear, 6 control of power-on up-shifts, 110–112
types of, 6–7 double-transition shifts, 120–121
see also Dual-clutch transmissions, Metal dynamic simulation model for studying, 110
pushing V-belt continuously variable gear-shifting mechanics, 105–110
transmissions, Planetary gear automatic hydraulic control system for, 110
transmissions system design, 112–114
transitional shifts, 119–120
Backward-searching strategy, 138–141 Clutch torque, 105
Brake specific fuel consumption (BSFC), 2–3 Clutch torque capacity, 105
Constant-output-torque power-on up-shift
Canceled shifts, 118–119 control, 109–110, 154–156
Centrifugal pendulum vibration absorber Continuously variable transmissions (CVTs), 6
(CPVA), 156–166 see also Metal pushing V-belt continuously
basic concept of, 157–158 variable transmissions
equations of motion of, 159–161 Coupling state, 12
201
6293_Book.indb 201 1/11/13 3:55 PM

Index
Double-transition shifts, 37, 120–121 synchronizer and its control, 187–189

Down-shift Filter-observer (FO) acceleration estimator,
fuzzy logic rules for shift points, 148–150 116–117
power-on down-shift, 42, 109–110 Fluid flow equation, 72
control of, 112 Friction launch control, 129–135
Driveline compliance 56 Frictional clutches, 24–25
Dry dual-clutch modules, 190–192 hyperbolic tangent function model, 45–46
Dual-clutch transmissions (DCTs), 6 mathematical models of, 45–47
construction of, 186 proportional and integral (PI) control
dynamics and controls of, 185–193 techniques, 46
generic models for, 57–60 Fuzzy controls, 145–148
matrix models for, 60–62 Fuzzy logic operations, 144
Duty cycle, 82 Fuzzy logic rules for shift points, 148–150
Dynamic modeling, 14 Fuzzy set, 143–144
Dynamic programming optimization, 137–138,
141–142 Gear ratio, 2, 5, 24, 186
Gear selection, power-based 152–154
ECCC mode, 122–123 Gear-shift control, clutch-to-clutch, 105–121
Efficiency, 12 Gear-shifting mechanics, 37–42
Eight-speed planetary automatic transmissions, from control perspective, 105–110
65–69 power-on down-shift, 42
Electrohydraulic pressure control system, 72–105 power-on up-shift, 38–41
hydraulic clutch control system analytical
study, 84–98 Hydraulic clutch control systems
PPC solenoid model, 80 analytical study of, 84–98
pressure control system and simulation models, block diagrams and transfer functions for,
73–79 87–90
pressure control valves, high-flow, direct- for clutch-to-clutch shift controls, 110
acting, 81 dynamic equations for, 85–87
pulse width modulated (PWM) solenoid, dynamic simulation models of, 91
82–84 system design, 91–98, 112–114
Electronic torque converter clutch control, for torque converter clutch control, 121
121–126 Hydraulic pistons, 76–77, 86–87
electronic control algorithm for, 122–123 Hydraulic pressure control system and simulation
hydraulic system for, 121 models, 73–79
End phase, 112 Hydraulic ratio control system, 171–173
Engine Hyperbolic tangent function clutch model,
dynamic model of, 14–18 45–46
engine power, 2
operation of, 5–6 Inertia balancing, 62–65
Engine loader control concept, 130–132 Inertial phase, 40, 42, 107–108, 111–112
Engine torque command generator, 151 Input K-factor, 12–13
Engine torque control system, 151–152 Input torque ratio, 28
Internal combustion engine, 2
Feed-forward/feedback control
application to V-CVT ratio control, 177–184 k, calculation of, 132–133
as direct pulley pressure control, 181–183 Kalman acceleration estimator, 114–116
control algorithms for, 192 kpki model, 46
control systems, 179–180
dual-clutch modules, 189–192 Lagrange equation
feed-forward control design, 180–181 for planetary gear trains, 49–52, 54–56
feedback controls, limitations of, 177–179 for simple planetary gear set, 47–49
202
6293_Book.indb 202 1/11/13 3:55 PM

Index
Lever analogy method for planetary gear trains, reduced-order equation for, 53
29–32 static analysis of
Lever diagram, 22 algebraic method, 26–29
Low-high-pressure ratio control system, 176 lever analogy method, 29–32
matrix method, 33–37
Master-slave ratio control system, 176–177 Power-based gear selection, 152–154
Matrix method for analysis of planetary gear Power-on down-shift, 42, 109–110
trains, 33–37 control of, 112
Membership function, 143–144 Power-on up-shift, 38–41, 106–109
Metal pushing V-belt continuously variable constant-output-torque control, 154–156
transmissions, 167–184 control of, 110–112
comparison with other ratio control systems, Powertrain, 2, 5
173–177 integrated control system for
controls of, 170–173 architecture of, 150–152
feed-forward/feedback control, 177–184 constant-output-torque power-on up-shift
mechanics of, 168–169 control, 154–156
power-based gear selection, 152–154
Off-going clutch control methods, 111 Pressure control valves, high-flow, direct-acting,
Orifice flow equation, 72 81
Output torque ratio, 28 Pressure regulation spool valve, 74–76, 85–86,
87–89
Pedal input interpreter, 150–151 design of, 91–94
Pipe flow equation, 73 Proportional pressure control (PPC) solenoid,
Planet carrier, 21–22 73–74, 76, 80
Planet gears, 21 Propulsion power curve, 4–5
Planetary gear automatic transmissions, 6 Pulse width modulated (PWM) solenoid, 82–84
centrifugal pendulum vibration absorber Pump speed, 2, 13, 18
(CPVA), 156–166 Pump torque, 12, 13, 18
clutch-to-clutch gear-shift control, 105–121
control of, 71–166 Ratio control systems, comparison of, 173–177
eight-speed, 65–69 Ratio thrust force, 170–171
electrohydraulic pressure control system, Reduced-order equation
72–105 for planetary gear trains, 53
electronic torque converter clutch control, for simple planetary gear set, 49
121–126 Regulated clutch fill control, 95
friction launch control, 129–135 Released mode, 122
integrated powertrain controls, 150–166 Releasing mode, 122
mechanics of, 10–69 Ring gear, 21–22
shift scheduling system, 135–150
six-speed, 65–69 Shift maps, 135–137
Planetary gear sets, 21–23 dynamic-programming-based generation of,
Lagrange equation for, 47–49 137–142
reduced-order equation for, 49 power-based, 152
simple, dynamic equations for, 47–49 Shift scheduling system, 135–150
Planetary gear trains, 21–37 artificial intelligence-based, 142–150
dual-clutch models, generic, 57–62 shift map, 135–137
dynamic equations for, 49–53 dynamic-programming-based generation of,
frictional clutches in,45–46 137–142
inertial balancing of, 62–65 Simulink models
Lagrange equation for, 49–52 centrifugal pendulum vibration absorber
obtained directly, 54–56 (CPVA), 161
mechanics of gear shifting, 37–42 dual-clutch system, 59
203
6293_Book.indb 203 1/11/13 3:55 PM

Index
Simulink models (cont.) Torque converter clutch (TCC), 10, 13–14

engine and torque converter system, 15 Torque converter clutch (TCC) control
engine loader friction launch control electronic, 121–126
algorithm, 131 hydraulic, 121
four-speed transmission, 51, 53 Torque converter clutch damper, dynamic
frictional clutch, 46 analysis of, 126–128
hydraulic clutch control system, 77 Torque coupling state, 12
pressure control system, 91 Torque multiplication state, 12
pump speed and pump torque, 18 Torque phase, 39–40, 106–107
torque converter clutch control algorithm, 123 control of, 111
Single-transition shift, 37 Torque ratio, 2, 12 13, 28, 29, 32, 35–37
Six-speed planetary automatic transmissions, Transitional shifts, 119–120
65–69 Transmissions
Solenoids necessity for, 2–6
proportional pressure control (PPC), 73–74, 76, connecting model to vehicle and engine, 56–57
80 see also Automatic transmissions, Dual-clutch
pulse width modulated (PWM), 82–84 transmissions, Metal pushing V-belt
Speed-change control method, 111–112 continuously variable transmissions,
Speed constraint equations Planetary gear automatic transmissions
for planetary gear sets, 22, 33 Trim valve, 94
for planetary gear train, 23–24, 27 Truth value, 143–144
Speed ratio, 2, 12, 26–27, 29, 30–31, 33–35 Turbine speed, 12, 13, 18
Step gear transmissions, 6 Turbine torque, 12, 13, 18
Stick diagram, 22
Sun gear, 21–22 Up-shift
Supply-side orifice, 95–96 fuzzy logic rules for shift points, 148–150
power-on up-shift, 38–41, 106–109
Tlau, calculation of, 133–134 constant-output-torque control, 154–156
Tloc, calculation of, 134 control of, 110–112
Torque capacity of clutch, 25, 38–39
Torque-capacity thrust force, 170 Vehicle drag load, 56
Torque converter
backward calculation of variables, 18–21 Wave plates, 96–98
description of, 11–14 Wet dual-clutch modules, 189
dynamic model of, 14–18 Wheel power, 4
204
6293_Book.indb 204 1/11/13 3:55 PM

About the Authors
Dr. Shushan Bai earned his BS degree in mechanical engineering

from Huazhong University of Science and Technologies, China, and
his MS and PhD degrees in electrical engineering from Hokkaido
University, Japan, both in the field of automatic controls. In his 23
years with GM Powertrain, he worked on advanced development
of automatic transmissions. He is an adjunct faculty member in the
mechanical engineering department at the University of Michigan. He has authored
numerous technical publications and holds many patents.
Joel M. Maguire received his graduate degree (MSOT) from Rens-

selaer Polytechnic Institute and his undergraduate degree from
Michigan Technological University. He began his automotive career
with General Motors in 1985 and currently is global innovation
leader in GM Powertrain’s advanced hybrid group. His work has
involved design, development, and analysis of automatic and manual
transmissions, as well as experimental transmission projects in North America and
Europe. He holds dozens of drivetrain-related patents.
Dr. Huei Peng received his PhD from the University of California,
Berkeley. He is a professor in the mechanical engineering depart-
ment at the University of Michigan. His research interests include
adaptive control and optimal control, with emphasis on their applica-
tions to vehicular and transportation systems. His current research
focuses include design and control of hybrid vehicles and vehicle
active safety systems. He has authored numerous technical papers and three books.
Dr. Peng received the National Science Foundation (NSF) Career award in 1998, and
is an ASME Fellow and Chang Jiang Scholar at Tsinghua University.
205
6293_Book.indb 205 1/11/13 3:55 PM

6293_Book.indb 206 1/11/13 3:55 PM


Dynamic Analysis and Control System Design of Automatic Transmissions

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dynamic Analysis and Control System Design of Automatic Transmissions

Uploaded by

Copyright:

Available Formats

Downloaded from SAE International by University of Michigan, Monday, October 29, 2018

Dynamic Analysis and

6293_Book.indb 1 1/11/13 3:51 PM

Other SAE books of interest:

Innovations in Automotive Transmission Engineering

Continuously Variable Transmission (CVT)

For more information or to order a book,

6293_Book.indb 2 1/11/13 3:51 PM

Dynamic Analysis and

Copyright © 2013 SAE International eISBN: 978-0-7680-7932-6

6293_Book.indb 3 1/11/13 3:51 PM

400 Commonwealth Drive

Copyright © 2013 SAE International. All rights reserved.

Library of Congress Cataloging-in-Publication Data

To purchase bulk quantities, please contact:

Visit the SAE International Bookstore at http://books.sae.org

6293_Book.indb 4 1/11/13 3:51 PM

Chapter 1 Automatic Transmissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Chapter 2 Mechanics of Planetary Gear Automatic Transmissions . . . . . . . . . . 9

6293_Book.indb 5 1/11/13 3:51 PM

Chapter 3 Control of Planetary Gear Automatic Transmissions . . . . . . . . . . . . 71

Chapter 4 Metal Pushing V-Belt Continuously Variable Transmissions . . . . . 167

6293_Book.indb 6 1/11/13 3:51 PM

4.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177

Chapter 5 Dynamics and Controls of Dual-Clutch Transmissions . . . . . . . . . . 185

6293_Book.indb 7 1/11/13 3:51 PM

6293_Book.indb 8 1/11/13 3:51 PM

6293_Book.indb 9 1/11/13 3:51 PM

6293_Book.indb 10 1/11/13 3:51 PM

6293_Book.indb 1 1/11/13 3:51 PM

An automatic transmission is a small component in automobiles, yet it plays an impor-

1.1 Powertrain system

1.2 Why is a transmission necessary?

Fig. 1.1 Powertrain

6293_Book.indb 2 1/11/13 3:51 PM

Fig. 1.2 Engine torque graph.

Fig. 1.3 Engine power graph.

Fig. 1.4 Engine BSFC graph.

6293_Book.indb 3 1/11/13 3:51 PM

1.2.1 First perspective: wheel power

Fig. 1.5 Propulsion power curve

Fig. 1.6 Propulsion power

6293_Book.indb 4 1/11/13 3:51 PM

Fig. 1.7 Propulsion power

1.2.2 Second perspective: engine operation

Fig. 1.8 Constant power

6293_Book.indb 5 1/11/13 3:51 PM

1.3 Different types of automatic transmissions

Fig. 1.9 Cross-section view of a planetary gear automatic transmission.

6293_Book.indb 6 1/11/13 3:51 PM

Fig. 1.10 Cross-section view of a continuously variable transmission.

Fig. 1.11 Cross-section view of a dual-clutch transmission.

6293_Book.indb 7 1/11/13 3:51 PM

6293_Book.indb 8 1/11/13 3:51 PM

6293_Book.indb 9 1/11/13 3:51 PM

A planetary gear automatic transmission typically consists of a torque converter, a plan-

Fig. 2.1 Conceptual diagram of a four-speed planetary gear automatic transmission.

6293_Book.indb 10 1/11/13 3:51 PM

Mechanics of Planetary Gear Automatic Transmissions

Fig. 2.2 Cross-section view of a planetary gear automatic transmission.

2.1 Torque converter

Fig. 2.3 Cross-section view of a torque converter.

6293_Book.indb 11 1/11/13 3:51 PM

Fig. 2.4 Conceptual diagram

Te − Tp − Ttcc = (I e + I p )ω e + D pωe (2.5)

Tt + Ttcc = (I t + I v )ω t + Dt ωt (2.9)

Tt_matrix =  Tt_indexT Tt_indexT