Development of a Clutch Assisted Engine Start
Politecnico di Torino
Porto Institutional Repository
[Doctoral thesis] Development of a Clutch Assisted Engine Start
Original Citation:
Gianmarco Brunetti (2014). Development of a Clutch Assisted Engine Start. PhD thesis
Availability:
This version is available at : http://porto.polito.it/2543152/ since: May 2014
Published version:
DOI:10.6092/polito/porto/2543152
Terms of use:
This article is made available under terms and conditions applicable to Open Access Policy Article ("Creative Commons: Attribution-No Derivative Works 3.0") , as described at http://porto.
polito.it/terms_and_conditions.html
Porto, the institutional repository of the Politecnico di Torino, is provided by the University Library
and the IT-Services. The aim is to enable open access to all the world. Please share with us how
this access benefits you. Your story matters.
(Article begins on next page)
POLITECNICO DI TORINO
SCUOLA DI DOTTORATO
Dottorato in Meccanica XXVI ciclo
Tesi di Dottorato
Development of a Clutch Assisted
Engine Start
Enabling Stop & Start Sailing on next generation powertrain
Gianmarco Brunetti
Tutore
Prof. Giovanni Belingardi
Coordinatore del corso di dottorato
Prof. Luigi Garibaldi
Marzo 2014
The contents of this document may not be reproduced in any form or communicated to any third party without the prior written consent of the author. While
every eort is made to ensure its correctness, the author assumes no responsibility
neither for errors and omissions which may occur in this document nor for damage
caused by them.
All rights reserved.
A papá
Indice
1 Optimizing HEV powertrain design to achieve CO2 targets (5)
1.1 Model denition . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1.1 Design Variables . . . . . . . . . . . . . . . . . . . . . . .
1.1.2 Design of Experiment . . . . . . . . . . . . . . . . . . . . .
1.1.3 Optimization . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Power-train design . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2.1 Battery sizing . . . . . . . . . . . . . . . . . . . . . . . . .
1.2.2 Power-train evolution scenario . . . . . . . . . . . . . . . .
1.2.3 Architecture selection . . . . . . . . . . . . . . . . . . . . .
2 Enabling sailing for next generation stop & start
2.1 Stop &Start . . . . . . . . . . . . . . . . . . . . . .
2.1.1 Upcoming regulatory framework . . . . . . .
2.1.2 Evolution of Stop & Start Technology . . . .
2.2 Sailing . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.1 Simulation tool . . . . . . . . . . . . . . . .
2.2.2 Simulation results . . . . . . . . . . . . . . .
2.2.3 CO2 gures . . . . . . . . . . . . . . . . . .
2.3 System Requirements . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3 Modeling of the vehicle driveline
3.1 Engine starting system . . . . . . . . . . . . . . . . . .
3.1.1 Starter motor . . . . . . . . . . . . . . . . . . .
3.1.2 Dual Mass Flywheel . . . . . . . . . . . . . . .
3.1.3 Engine . . . . . . . . . . . . . . . . . . . . . . .
3.1.4 Model validation . . . . . . . . . . . . . . . . .
3.2 Transmission . . . . . . . . . . . . . . . . . . . . . . .
3.2.1 Clutch . . . . . . . . . . . . . . . . . . . . . . .
3.2.2 Gearbox, dierential and nal drive . . . . . . .
3.2.3 Wheels and vehicle . . . . . . . . . . . . . . . .
3.2.4 Validating modeling of transmission components
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
v
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
2
3
11
14
14
15
17
19
21
22
22
23
27
27
29
30
32
33
34
35
36
46
55
59
61
66
67
70
3.3 Analysis of vibrations of the drive-line . . . . . . . . . . . . . . . . . 71
4 Clutch Assisted Start
4.1 Development of clutch control . .
4.1.1 Predictive reduced model .
4.1.2 Main clutch operation . .
4.1.3 Torque control . . . . . .
4.1.4 Simulation results . . . . .
4.2 Optimization of clutch control . .
4.2.1 Criteria & Requirements .
4.2.2 Methodology . . . . . . .
4.3 Results . . . . . . . . . . . . . . .
4.4 Conclusions . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
A In-cylinder pressure predictive model
A.1 Predictive Combustion Model . . . .
A.2 In-cylinder pressure generation . . . .
A.3 Multi-cylinder pressure model . . . .
A.4 Co-simulation between AMESim &
MATLAB/Simulink . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
81
82
82
84
86
87
88
90
91
97
102
103
. . . . . . . . . . . . . . . . . . 103
. . . . . . . . . . . . . . . . . . 105
. . . . . . . . . . . . . . . . . . 108
. . . . . . . . . . . . . . . . . . 109
B AVL-Drive (21)
113
B.1 System Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
C ACRONYMS
117
vi
Aknowledgements
The research has been conducted within General Motors Powertrain - Europe. I
would like to thank Ing. Maurizio Cisternino, my company tutor, for supervising
the work and for strongly believing in the outcomes of the project.
My deep appreciation and regard to Prof. Giovanni Belingardi for supervising
my Ph.D program and acting as mentor both for my life and career.
1
Abstract
Environmental protection and ecient energy utilization have been always important issues in the automotive industry, but have gained signicant momentum with
the growing demand for mobility around the world and its impact on the global
environment. For this purpose, many improvements in automobile technology have
been accomplished over the past decades. However fuel economy with improvements
in vehicle powertrain technology has been penalized by customer preferences. Automotive industry faces the challenge of producing vehicles that meet future fuel
economy and emissions requirements which are priced to meet the desired customer
value.
As hybrid vehicles, due to the high cost of the electrication they introduce, in
next years will not impact the OEM eet-averaged CO2 gures in a signicant way,
it is benecial to introduce new cost-oriented CO2 features able to optimize engine
operations, as they oer a very favorable cost/benet ratio.
According to market trend, the increasing interest on automated transmission
(i.e. AMT, DCT) plays a key-role towards the optimization of engine operation.
The basic principle of shutting the engine o at idle to remove engineâs drag torque could be adopted at vehicle in motion, extending the distance covered by the
vehicle coasting, when no traction is required, by opening the clutch automatically.
Literature calls such operation âsailingâ: represents a low cost control feature, as it
does not introduce new components, able to enhance Start & Stop technology.
Turning o the combustion engine during coasting conditions increases the number of engine starts over vehicle lifetime, impacting the whole vehicle.
Today starter motors are typically designed for Start&Stop applications for up to
300.000 engine crank over vehicle lifetime. This number can double for Start&Stop
Sailing applications. However, the number of load cycles for the starter motor can
be reduced signicantly if the engine is revved up via clutch.
The work focuses on the development clutch assisted start by applying a new
methodology based on modeling and simulation.
In general, starting the engine through clutch engagement consists in spending a
certain amount of the traveling vehicle kinetic energy to spin up the engine. The consequent deceleration of the vehicle, due to this maneuver, must be mitigated by an
3
appropriate clutch closing strategy. Moreover, vibrations induced in the powertrain
during the maneuver can generate uncomfortable feelings for the passenger.
Hence a powertrain and driveline modeling that allows to investigate the impact
on drivability for dierent clutch engagement in dierent operating conditions in
order to fulll drivability requirements.
4
Capitolo 1
Optimizing HEV powertrain design
to achieve CO2 targets (5)
In recent years the pursuit of extremely low CO2 emissions has gathered interest by
OEM due to a combination of legislative and market requirements. In particular
the European Commission mandate for 95 gCO2 /km in 2020, as well as a set of
market incentives, which favor fuel ecient vehicles, is creating a strong request for
technical solutions that may be applied in the OEMs portfolio in order to rapidly
impact the eet overall emission.
Hybrid Electric Vehicles (HEVs) can be considered one of the most promising
ways of improving the sustainability of the road transport sector. They are equipped
with an Internal Combustion Engine (ICE) coupled to an electro-mechanical system.
Legislation oers an opportunity in developing parallel-hybrid powertrain featuring
a direct electric drive and, optionally, an o-vehicle charging disposal. This layout
allows to perform pure electric range, achieving a benet in the computation of fuel
economy (ECE Regulation No. 101).
Although Hybrid Electric Vehicles (HEVs) can be considered a viable technology to transition from conventional mobility to electric-mobility, their diusion is
still limited due to high costs. Technologies, in general, in order to become massproduction products should meet cost requirements, including the integration on
vehicle. Automotive industry faces the challenge of producing vehicles that meet
future fuel economy and emissions requirements which are priced to meet the desired customer value. Therefore, this chapter will present a methodology to optimize
the design of a HEV powertrain, of its electric components and the vehicle mass in
order to achieve CO2 emission targets by using a meta modeling approach. Starting from a base diesel powered architecture, parameters focus of the optimization
include base vehicle curb mass, battery size, degree of hybridization. For all architectures, selected by a DoE approach, calculated results includes additional cost
to base architecture, CO2 (according to EU norms), distance performed in pure
1
1 Optimizing HEV powertrain design to achieve CO2 targets (5)
Figura 1.1: Layout of the considered hybrid powertrain
electric, hybrid functionalities (i.e. pure electric mode, regenerative braking, AWD
mode). Parameters optimization is performed by empirical target function, based
on gathered results.
After model denition, optimization investigates the most ecient way to achieve
given fuel economy targets, expected functionalities, minimizing the additional cost
of the architecture. An ecient way to achieve fuel economy targets for a mid-size
class vehicle is proposed as the best compromise between all the analyzed targets.
Final investigation evaluate the impact of adoption of lighter materials for BIW on
the achievement of CO2 targets.
1.1 Model denition
Increasingly stringent requirements for lower CO2 emissions and the demand for a
diversication of energy carriers are driving the development of future vehicles. In
mid to longer term, evolutionary propulsion technologies with signicant numbers in
global vehicle eet will impact on global fuel consumption and emission reductions.
Emerging technologies such as electrication, in combination with optimized ICE
have proved that the reductions in fuel consumption and emissions could be attained.
However, the higher cost of the technology than traditional powertrain is limiting
the impact on the eet-averaged CO2 gures in a signicant way.
Among conventional powertrain, diesel engine are naturally more fuel ecient
than gasoline, as reported in (3). Despite this advantage, it is becoming dicult to
meet the increasingly stringent emission standards in a cost-ecient way. Therefore,
2
1.1 Model denition
the analysis will focus on diesel powertrain as the higher diusion in the European
market in the mid-size and mid-size compact segments impacts the eet-avereged
CO2 gures signicantly and as their lower fuel economy and their higher production
cost makes diesel a worst case for hybridization, reducing space for improvement by
electrication (i.e. lower FE improvement, higher cost of the powertrain).
The study adopts a metamodeling design approach, allowing the comparison of
dierent powertrain technologies, electrication architectures and BIW materials.
There are a few signicant dierences between conventional and hybrid electric vehicles in terms of their powertrain components. Firstly, hybrid powertrains require
additional components, such as electric motors, power electronic devices, and highvoltage battery packs. Secondly, they typically need modied design for the existing
components (e.g. a smaller engine) because they have a new set of components. Lastly, some of the hybrid powertrains adopt novel congurations that utilize their
components in a considerably dierent way from conventional ways. All these changes have to be carefully managed for the vehicles to be able to achieve performance
that is close to their potential.
The purposes of employing metamodels in this study are manifold. It enables the
adoption of numerical optimization techniques without using scaling of a baseline
design. One favoured way in architectural studies is picking the best design othe-shelf. This can refer to selecting among either dierent technologies or dierent
size levels. Optimization techniques are essential in order to explore the vast design
space with respect to engine and battery size as well as powertrain congurations
and use light-weight materials for BIW. The optimization formulations includes cost
for components and set penalty on emissions.
The metamodels developed are based on a mathematical, statistical approach
with the aim of providing performance information of in-between designs, thus making it possible to use continuous design variables in the optimizations. In order
to select candidates for a model set-up, the study adopts a DoE approach: candidates selection has been performed via optimal designs, allowing parameters to
be estimated without bias and with minimum-variance; the criterion adopted is
V −optimality, based on prediction variance, which seeks to minimizes the average
prediction variance over a set of n specic points.
1.1.1 Design Variables
The rst step towards optimization of the design of an HEV architecture is to dene
the main targets of the design.
HEV type The state-of-art technology of HEVs has brought to the market diverse
hybrid architectures: literature has dened a main criterion to list architecture in 3
main categories according to the source of energy providing motion to the vehicle:
3
1 Optimizing HEV powertrain design to achieve CO2 targets (5)
ˆ series HEV â only EM drives the vehicle
ˆ parallel HEV â both EM and ICE drive the vehicle
ˆ seriesparallel HEV â allowing both previous layout
Both literature and market data have proven that a parallel HEV architecture
represent the most promising technology for light-duty passenger cars. A suitable
parameterization to investigate architecture accounts for the type of parallel HEV
focusing on position of the EM in the powertrain. Tabella1.2 reports the value of
the HEV type index according to the layout, described in Figura1.2.
a ) P1 powertrain layout
b) P 2 powertrain layout
c ) P 3 powertrain layout
d) P 4 powertrain layout
Figura 1.2: Hybrid powertrain layout according to the type of parallel HEV.
Index
1
2
3
4
HEV
type
Parallel
Parallel
SeriesParallel
Parallel
HEV
layout
EM linked
to ICE
EM linked
to gearbox
EM combined
in transmission
EM linked to
separate driving axle
Tabella 1.1: The HEV powertrain layout
DoH
Mild HEV
Full HEV
Plug-in HEV
EREV
Electric
Power
Low
Med
Med
High
Electric
Storage
Very low
Low
Med
High
Grid
Recharge
No
No
Yes
Yes
Pure
EV
0%
10%
30%
70%
Tabella 1.2: Attributes of Electrical Vehicles (1)
4
Drive
Range Limit
Fuel
Fuel
Fuel
+ Battery
Fuel
+ Battery
1.1 Model denition
Degree of Hybridization Literature calls the increase in electrical content and
magnitude onto the vehicle âelectricationâ, indicating the development and integration of systems and components that enable electric energy to be used for
transportation.
Hybrids have been traditionally classied by the amount of driving power supplied by the electrical system and the amount supplied by the engine. HEVs with
large electrical systems and very small engines, this denition works pretty well.
It also works relatively well for vehicles that do not have a downsized engine and
have simply added on a technology referred to as an integrated starter generator:
these are just conventional vehicles that can turn the engine o when the vehicle is
stopped.
Both the amount of electrication and features supplied by electrication dene
the degree of hybridization on board of the vehicle.
60%
Plug-in charging
High EV Drive Plug-in charging
Regen
Extended EV Drive
E-Assist
Regen
S&S
E-Assist
S&S
50%
Electrification
40%
Low EV mode
Regen
E-Assist
S&S
30%
Regen
E-Assist
S&S
20%
10%
S&S
0%
Conventional
Micro
Mild HEV
Full HEV
Plug-in HEV
EREV
Figura 1.3: The amount of electrication at each degree of hybridization.
Battery size Lithium-ion batteries have become the standard for electric vehicles
that store energy from the grid because of their superior energy density compared
to other technologies. Batteries are mainly controlled by their SOC: in order to
preserve their state of health through operations, they are usually limited in charging/discharging, therefore limiting the energy storage capabilities. Considered size
of batteries and range of use of SOC are reported in Tabella1.3.
5
1 Optimizing HEV powertrain design to achieve CO2 targets (5)
Degree of
Hybridization
S&S
Mild
Full HEV
Plugin HEV
EREV
Energy
kWh
SOC range
%
0
0.5
2
4
16
0
0.3
0.5
0.5
0.5
Tabella 1.3: Energy content, usable SOC range of batteries according to the degree
of hybridization of the powertrain
Vehicle size The goal of the methodology is to optimize cost of hybridization for
a specic vehicle, therefore the model includes the size of the vehicle as a design
parameter/control parameter.
The analysis refers to vehicle in major segments for European market Tabella1.4
in their baseline diesel powered conguration. Mass for type approval certication,
so called Inertia Weight Class is reported as well.
Segment
Size
B
C
D
IWC
kg
1
2
3
1250
1360
1470
Base
powertrain
Small Diesel
Mid-Sized Diesel
Large Diesel
Tabella 1.4: Vehicle segmentation
Displacement Hybridizing a diesel powertrain provides signicant fuel economy
improvements over a conventional gasoline vehicle. However, these fuel economy
benets are subject to emissions, cost and degree of hybridization. Therefore, optimizing the design of a diesel-hybrid vehicle represents a greater challenge and
assumes higher signicance when looking at the European market.
Electrication oers opportunities for engine downsizing according to the degree
of hybridization. The engine displacement describes the size of an engine. Moreover,
according OEMsâ trend to reduce manufacturing costs by adoption of equal cylinder
set (i.e. bore and stroke) for any ICE, downsizing is performed by reduction of the
number of cylinders.
All powertrain considered must full EU 6 emission levels, by adopting aftertreatment disposals. Tabella1.5 reports engines evaluated in the study.
6
1.1 Model denition
Size
Cylinders
X-Small
Small
Mid
Large
Power
kW
Displacement
cm3]
2
3
4
4
45
70
95
115
800
1200
1600
2000
Tabella 1.5: Main data of ICE focus of the analysis
ICE contents Due to the high cost of electrication, OEMs lately tend to optimize
baseline engine by introducing new contents to reduce diesel emissions and engine
frictions, to enhance performance and warm-up.
A proper design of the powertrain towards CO2 emissions reduction considers
the introduction of ICE contents, in order to quickly compare electrication to conventional ICE improvement. Only for the purpose of the study 2 dierent contents
have been analyzed: one reducing engine friction, one enhancing performarce. The
model can consider any specic ICE content for evaluation.
Mass Reduction A major point of the analysis is to evaluate the impact of a
mass reduction of BIW, by introduction of new materials and technologies for BIW
(12).
Though enhancing energetic eciency of powertrain, electrication implies a
mass increase of the vehicle due to electric components ( i.e. motors, batteries),
leading to an higher energy request of the vehicle to complete a certain mission
prole. As showed in 1.1, mass (m) impacts the the power required (Preq ) by the
vehicle during accelerations phases (Pacc ). Resistive power (Pres ) is calculated by
use of coast-down coecients.
Preq = Pres + Pacc
Pres = (F0 + F1 · v + F2 · v 2 ) · v
·v
Pacc = m · ∆v
∆t
(1.1)
In the design of electric components, i.e. battery, a proper trade-o between
vehicle mass increase and battery size is dened in order to achieve CO2 target,
distance in pure EV mode.
The introduction of a light-weight design both mitigates the increase of vehicle
mass due to electrication and leads to lower CO2 values that might signicantly
impact eet-averaged CO2 gures.
Design Output
7
1 Optimizing HEV powertrain design to achieve CO2 targets (5)
Figura 1.4: Impact on costs of technologies for mass reduction of BIW.
CO2 emissions The main target for hybridization is the reduction of CO2 emissions both to deliver better fuel economy .to customers and to achieve regulated
emission compliancy.
Fuel economy for any selected architecture has been evaluated via simulation12 .
1 Models for HEV powertrain include an optimization logic, so called Hybrid Operating Strategy
(HOS): it provides for determining an overall value of mechanical power to be delivered to the
wheels of the motor vehicle, for splitting this overall value in a rst contributing value of mechanical
power to be requested to the ICE and a second contributing value of mechanical power to be
requested to the electric motors. In greater details, the splitting of the above mentioned overall
power value is determined by minimizing the power losses due to the operation of the hybrid
powertrain. Power losses are accounted in a predetermined polynomial function, usually referred
as target function, which quanties a quantity of power supplied to the hybrid powertrain through
the ICE fuel, but not delivered to the wheels(3).
2 CO gures for Hybrid vehicles are calculated according to ECE Regulation No. 101, therefore
2
measured on type approval homologation cycle. For PHEV, EREV only CO2 value is evaluated by
averaging the emissions of a cycles performed with fully-charged battery with the ones performed
1 +Dav ·M2
at minimum SOC. Then each emission value is weighted as follows: M = De ·M
, where De
De +Dav
is the distance performed in pure EV mode and Dav is an average value between 2 charges set by
regulations to 25km.
8
1.1 Model denition
16
0.12
14
0.1
0.08
10
8
0.06
6
Energy [kWh]
Power [kW]
12
0.04
4
0.02
2
0
0
0
50
100
150
200
time [s]
IWC 1130
IWC1250
IWC 1360
vehicle speed
Figura 1.5: Impact of vehicle mass on energy, power required over NEDC
homologation cycle.
Modelling and simulation of Fuel Economy, emissions The model is
based on a âquasi-staticâ approach: even though the fuel consumption and emissions are calculated as functions of the dynamic ICE speed and load proles, each
condition is based on experimentally-derived stationary maps.
The model was assessed by means of experimental tests carried out at the dynamic test bed and at the roller test-bench by means of prototype vehicles, at GMPT-E
test facilities.
Mass The vehicle mass represents a main driver towards the reduction of CO2
emission. Electrication increases the mass of the vehicle, lowering, therefore,
the fuel economy improvements. Though, the introduction of new technologies
and materials for BIW oers a good opportunity to contain mass increase due to
electrication and battery size for electrication
Performance Performance are represented by an index of adequacy for the powertrain design for a given platform. Starting from an architecture 0, represented by the
proper base engine match for vehicle size Tabella1.6, an engine upgrade/downgrade
increases the index by ±1 (Figura1.6).
9
1 Optimizing HEV powertrain design to achieve CO2 targets (5)
Vehicle
Size
B
C
D
Displacement
[cm3
Power
kW
Performance
1200
1600
2000
70
95
115
0
0
0
Tabella 1.6: Architecture 0 for balanced performance.
Performance index is aected by mass reduction, battery size, engine contents,
type of hybridization.
2
Perofromance index
1
Architecture 0
0
Downsizing
Upsizing
-1
-2
-3
800
1200
1600
2000
Displacement [cm3]
Figura 1.6: Index of performance: impact of engine displacement.
Functionalities Any vehicle should full to certain requirements or expectation
from the customer. Electrication, according to the selected architecture, oers the
opportunity to develop new functionalities for the vehicles. The study considers:
1. Stop & Start - engine shut-o when no traction is requested
2. Electric boost â electric assistance to ICE
3. Pure EV mode
10
1.1 Model denition
4. All wheel drive
Index
Vale
1
2
3
4
Functionalities
S&S
S&S, Electric boost
S&S, Electric boost, Pure EV
S&S, Electric boost, Pure EV, AWD
Tabella 1.7: The functionality index.
Cost It accounts for the cost of the whole electrication and materials for BIW,
compared to the same vehicle in the baseline conguration. As goal of the study is to
achieve emissions compliancy in the most cost ecient way for OEMs, this parameter
is the one minimized according to given constraints. All design parameters impact
cost in dierent ways (Figura1.7).
Electric Range Besides contributing in the computation of CO2 emissions for
HEVs, the increasing focus on electric mobility, more stringent regulations to access
city-centre in major cities, make electric range an important target for HEVs. The
distance to be covered in a pure EV mode represents one of main criteria in the
design of vehicles for future mobility. The study evaluates through simulations the
electric range according to type approval procedures described in ECE Regulation
No. 101 (4).
Electric range mainly depends on the size of batteries, i.e. the amount of energy
stored, for a given hybrid architecture.
1.1.2 Design of Experiment
A full factorial combination of all design parameter leads to a large amount of
combinations to evaluate via simulation. The DoE seems the right approach as it
reduce the number of simulations run, considering both combination of eects and
eliminating unfeasible combinations of design parameters for instance full electried
vehicles with small batteries.
Matlab-based âModel based calibrationâ tool has been adopted to design experiments, limiting the space of variation of parameters, by introduction of specic
constraints. For instance, Figura1.8 shows that vehicles might introduce further
engine downsizing only at high degree of hybridization.
11
1 Optimizing HEV powertrain design to achieve CO2 targets (5)
Figura 1.7: Cost increase according to vehicle design parameters.
Figura 1.8: 4D projection of experiments for the design.
The model built from test results uses second grade polynomial functions 1.2. It
results that design output are linearly correlated to input design parameters.
12
1.1 Model denition
(CO
2 , Cost, F unctionality, P erf ormance, M ass, EV range) =
2
f HEVtype
, DOH 2 , % mass reduction2 , Size2ICE , Size2veh , ICEcontent2 , Sizebatt
(1.2)
Statistical parameters, used to check the quality of the model, show the high
correlation of predicted results versus simulation results.
Predicted/Observed
140
120
Predicted CO2 [[g/km]]
100
80
60
40
20
20
40
60
80
CO2 [[g/km ]]
100
120
140
Figura 1.9: Compare of predicted values and observed values of CO2 .
Model
CO2
Cost
Functionality
Performance
Mass
EV Range
R2
PRESS R2
0.937
0.981
0.971
0.996
0.841
0.851
0.879
0.964
0.949
0.983
0.695
0.739
Tabella 1.8: DoE model results.
13
1 Optimizing HEV powertrain design to achieve CO2 targets (5)
1.1.3 Optimization
The hybrid vehicle design methodology uses optimization techniques based on specic algorithms. Once models for design output are dened, a criteria to optimize
design parameters is decided.
According to the typology of this study, optimization adopts a MatLab based
âfminconâ is a function included in MatLab Optimization Toolbox © which nds
a constrained minimum of a scalar function of several variables starting at an initial
estimate. More specically nds the minimum of a problem specied in 1.3 by

C (x) ≤ 0




 ceq (x) = 0
A·x≤b
minx f (x) →
(1.3)


A
·
x
≤
b

eq
eq


lb ≤ x ≤ ub
x, b, beq , lb, and ub are vectors, A and Aeq are matrices, c(x) and ceq(x) are functions
that return vectors, and f (x) is a function that returns a scalar. f (x), c(x), and
ceq(x) can be nonlinear functions.
As underlined in (7), âthe powertrain system characteristics are highly nonlinear and non-continuous that may have a large number of local optimaâ. The SQP
algorithm (9) is an iterative method for nonlinear optimization, that takes every
iterative step in the region constrained by bounds. SQP methods are used on problems for which the objective function and the constraints are twice continuously
dierentiable.
Finally, based on above mentioned algorithms, the optimization is conducted
in a specic user-endly interface, allowing the user to select target function for
optimization, constraints and range of variation for design parameters.
1.2 Power-train design
OEMs, lately, faces the major challenge of designing fuel ecient powertrain which
are priced to meet the desired customer value.
In mid to longer term, evolutionary propulsion technologies (Figura1.11) with
signicant numbers in global vehicle eet will impact on global fuel consumption
and emission reductions. Emerging technologies such as electrication, in combination with optimized ICE have proved that the reductions in fuel consumption and
emissions could be attained. However the main challenge today is nding the right
mix of technical solutions or design parameters in order to achieve CO2 reduction
and fulll requirements at minimum cost.
14
1.2 Power-train design
Figura 1.10: The optimization tool interface.
For such matter, an optimization algorythm as presented in ?? is useful to draft
scenarios for powertrain development according to predened targets (i.e. cost, fuel
economy).
The methodology, therefore, has been applied to a specic engineering case, to
optimize the design of a mid-sized passenger vehicle. The compact vehicle must
achieve low fuel consumption over urban driving and must be capable of high fuel
economy over extra-urban driving at minimal cost. Therefore, the baseline for the
analysis is a conventional diesel powered vehicle.
1.2.1 Battery sizing
Before optimizing the design of the power-train according to requirements, it is
important the inuence of main design parameters on the design output by using
the available model.
The energy size of the battery is one of the major design parameter of the
powertrain as it inuences the CO2 emissions, the vehicle mass and it is the most
15
1 Optimizing HEV powertrain design to achieve CO2 targets (5)
Figura 1.11: Available technical solution to achieve fuel consumption reduction courtesy of Ricardo: GM workshop May 2012.
cost sensitive. Figura1.12, 1.13, 1.14 reect results of this analysis.
85
Plug-in
Plug.in
55
Plug-in
60
Plug-in
Plug-in
0.3
65
PLug-in
0
Full
70
Mild
CO2 [g/km]
75
Stop & Start
80
8
16
50
45
0.5
0.5
1
2
4
Battery size [kWh]
Figura 1.12: Impact of battery size on CO2 gures.
16
Plug.in
80
70
Plug-in
60
50
0
0.3
0.5
0.5
1
Plug-in
Plug-in
0
PLug-in
10
Full
20
Mild
30
2
Plug-in
40
Stop & Start
Driving range in pure electric mode real city driving [km]
1.2 Power-train design
4
8
16
Battery size [kWh]
Figura 1.13: Impact of battery size on Pure EV driving..
Overall, the methodology to optimize power-train design performed well. Thus,
for early-design stages of the vehicle development process, due to the high spread of
technologies and solutions leading to CO2 reduction, selecting the most convenient
path becomes a critical issue.
The adoption of mathematical approach towards optimization of the design at
early stages of development, can summarize eects of dierent technologies and
evaluate them on a fair mathematical/statistical base, optimizing towards a selected
target.
1.2.2 Power-train evolution scenario
The model developed can be further used to dene a technology road map for
electrication of diesel engines.
When looking at power-train electrication, two major scenarios might be drafted:
ˆ
Entry-level electrication applies to technical solution which might oer CO2
advantages at minimum cost;
17
Plug.in
1 Optimizing HEV powertrain design to achieve CO2 targets (5)
1400
Plug-in
Plug-in
1300
4
8
0.3
0.5
Plug-in
1050
Plug-in
1100
0.5
1
2
Stop & Start
1150
Full
1200
PLug-in
1250
Mild
Weight [kg]
1350
1000
0
Battery size [kWh]
16
Figura 1.14: Impact of battery size on vehicle mass.
ˆ
High-end electrication applies to technical solution which might oer signicant CO2 savings and customer features at high cost.
Entry-level electrication An optimized design for entry-level electrication
includes a P1 power-train design, having a small electric motor coupled to the engine.
A micro-hybrid electrication would be preferred oering a only Stop&Start feature,
which shuts the engine o when no traction is required.
High-end electrication An optimized design for high-end electrication includes either a P3 or a P4 power-train design, having a bigger electric motor coupled to
the transmission or on the rear axle. A full-hybrid electrication would be preferred
oering, additionally to Stop&Start, features a electric boost, energy recovery full
electric operation and perhaps All-Wheel-Drive, by running only with the electric
motor on the rear axle.
As in Figura1.15, by combining the 2 scenarios a technology road map can be
dened, where the diverse technical solutions for electrication are classied by cost
and CO2 saving potential.
18
1.2 Power-train design
Figura 1.15: Technology road map for electrication of diesel engines.
1.2.3 Architecture selection
In recent years the pursuit of extremely low CO2 emissions has gathered interest by
OEM due to a combination of legislative and market requirements. In particular
the European Commission mandate for 95gCO2 /km in 2020, as well as a set of
market incentives, which favor fuel ecient vehicles, is creating a strong request for
technical solutions that may be applied in the OEMs portfolio in order to rapidly
impact the eet overall emission.
Therefore, over the powertrain architecture selected over this study will apply
to entry-level electrication scenario, focusing on the development of micro-hybrid
vehicles, applying Stop & Start technology.
Other power-train architecture, in order to deploy the additional features, require
bigger batteries, which increase signicantly the overall cost of the power-train.
19
Capitolo 2
Enabling sailing for next generation
stop & start
Environmental protection and ecient energy utilization have been always important issues in the automotive industry Chapter 1, but have gained signicant momentum with the growing demand for mobility around the world and its impact
on the global environment. Towards this end, many improvements in automobile
technology have been accomplished over the past decades. However, fuel economy
with improvements in vehicle, powertrain technology have been penalized by customer preferences. Automotive industry faces the challenge of producing vehicles that
meet future fuel economy and emissions requirements which are priced to meet the
desired customer value.
As hybrid vehicles, due to the high cost of the electrication they introduce, in
next years will not impact the OEM eet-averaged CO2 gures in a signicant way,
it is benecial to introduce new cost-oriented CO2 features able to optimize engine
operations, as they oer a very favourable cost/benet ratio.
According to market trend, the increasing interest on automated transmission
(i.e. MTA,DCT, e-Clutch) plays a key-role towards the optimization of engine operation. The basic principle of shutting the engine o at idle to remove engineâs drag
torque could be adopted at vehicle in motion, extending the distance covered by
the vehicle rolling in neutral, when no traction is required, by opening the clutch
automatically. Literature calls such operation sailing: represents a low cost control
feature, as it does not introduce new components, able to enhance Start & Stop
technology.
The paper will assess the impact on fuel economy of opportune strategies to
enable sailing both over real world driving and on relevant regulatory schedules.
The study will focus on a Diesel engine powertrain, as its higher eciency than
gasoline reduces the space for improvement, it oers greater opportunity by removing
21
2 Enabling sailing for next generation stop & start
Speed
engineâs drag to extend ICE shut-o periods and, due to its higher diusion in the
European market, it impacts CO2 gures in signicantly.
Fuel Cufoff
Sailing
Distance
a)
b)
Figura 2.1: a) Engine Stop at vehicle in motion, b) Distance covered by vehicle with
std S&S control strategy and vehicle adopting sailing during coasting.
2.1 Stop &Start
Stop & Start technology improves fuel economy by reducing fuel used when the
vehicle is stopped, by stopping the engine rotation. It was introduced in Europe in
2008 equipping only a few population of vehicles; in 2012 more than half of vehicles
sold in Europe included such system. (17) This growth rate is due to a favourable
costbenet ratio, to improve fuel economy in the New European Driving Cycle (NEDC). Some additional capacity or durability additional components may be added
to the powertrain, ie; AGM battery and starter, and Stop & Start mainly requires
the development and renement of controls. Further fuel economy improvement
may be possible by extending Stop & Start operation to other vehicle speed ranges.
Though initially Stop & Start technology was primarily introduced in vehicles with manual transmission, this technology has been further developed to be applicable
to automated transmissions and show further CO2 benet on manual transmission.
Today, Start&Stop systems are implemented in all vehicle segments and markets,
from entry level up to the luxury segment.
2.1.1 Upcoming regulatory framework
Though, the main assumption behind this scenario is the NEDC cycle adopted for
dening fuel economy gures. Such homologation procedure will phase out starting in 2017 through 2020, being replaced by the World Ligh-Duty Test Procedure
22
2.1 Stop &Start
(WLTP), impacting, then, the scenario drafted till now.
Distance
Duration
Duration of stop phases
Maximum speed
Average speed w/o stops
Average speed w/ stops
Minimum acceleration
Maximum acceleration
Average acceleration
Average deceleration
Energy required over cycle
Average energy over cycle
[km]
[s]
[%]
[km/h]
[km/h]
[km/h]
[m/s2]
[m/s2]
[m/s2]
[m/s2]
[kWh]
[Wh/km]
NEDC
WLTP
11.02
1180
24.8
120
44.6
43.6
3.80
−5.00
2.14
−2.84
1.25
114
23.26
1800
13.4
131.3
53.8
46.5
6.00
−5.40
1.46
−1.6
2.44
105
Tabella 2.1: Key parameters of the NEDC, WLTP
In 2.1 the key parameters of both cycles are reported in order to compare NEDC
and WLTP. While in NEDC the vehicle spends almost a quarter of the time at idle,
over WLTP the vehicle is stopped only for about 13% of the time, impacting, then,
fuel economy benet of Stop & Start technology as it is intended today. Although
WLTP cycle shows stronger acceleration, deceleration than NEDC, the average values of acceleration, deceleration of WLTP are lower than NEDC (16). As a matter
of facts, several phases over WLTP are characterized by smooth deceleration proles Figure 2 b) below the 1m/s2 . Indeed, over NEDC, decelerations occurs less
frequently and show steeper proles.
2.1.2 Evolution of Stop & Start Technology
The basic principle of Stop & Start technology is the elimination of engine drag
torque at idling conditions, i.e. when propulsion is not required (i.e. at vehicle
stop). We will refer to Conventional Stop & Start in the following.By turning o
the combustion engine when decoupled from the drivetrain, the fuel that is injected
simply to overcome the combustion engine friction, is saved. A similar concept can be
applied in order to improve fuel economy at higher vehicle speeds. Nowadays, almost
on every vehicle all engine are equipped with smart engine controls as DFCO which
turns o fuel injection as soon as the driver releases the gas pedal. Though, during
DFCO, engine is driven by the vehicle, its friction and pump losses cause a signicant
loss of energy. Depending on the engine size, type and design, the motoring torque
and therefore the resulting motoring power can be signicant. Figura2.3 reports
23
2 Enabling sailing for next generation stop & start
Speed, Desired vehicle (km/h)
a)
150
Speed NEDC
Speed WLTP v5.3
100
50
0
0
200
400
600
800
1000
1200
1400
1600
1800
1200
1400
1600
1800
Time (sec)
Acceleration, Desired vehicle (m/s2)
b)
10
Acceleration WLTP v5.3
Acceleration NEDC
5
0
-5
-10
0
200
400
600
800
1000
Time (sec)
Figura 2.2: Speed (a), acceleration (b) proles of NEDC and WLTP
typical values of engine fmep (friction mean eective torque) for turbocharged lightduty diesel engine, depending on engine speed. The engine speed in a vehicle depends
on vehicle speed and the gear/transmission ratio.
4.5
4.0
fmep [bar]
3.5
3.0
2.5
2.0
1.5
1.0
0.5
Small
Mid
Large
0.0
0
1000
2000
3000
4000
5000
6000
Engine Speed [rpm]
Figura 2.3: Engine fmep for dierent sizes of conventional light duty turbocharged
diesels.
The fuel consumption benet provided by Stop & Start lately has been further
enhanced extending its range of applicability on cycle. Conventional Stop & Start
24
2.1 Stop &Start
system enables engine shut-o when vehicle is shifted into neutral at a stop. Due to
homologation procedures, such operation provides fuel economy improvements on
NEDC (2.4).
2
2
Clutch2
Clutch2
10 km/h
1
Neutral
Clutch1
Idle time
3s
21 s
Figura 2.4: Fuel economy opportunity during idle phases over NEDC
Although the energy required to motor the engine varies according to the vehicle
mission, the fuel engine propels its amount is signicant only might be quite signicant. Though, over the cycle the fuel provided is spent for more than 22% to motor
the engine (2.5).
To remove the engine motoring torque during coasting, a free-wheel overrunning
one-way clutch (OWC) may be included into the vehicle's transmission, allowing
the transfer of positive torque only. A OWC adds complexity of a second controlled bypassing clutch, for use during down-hill driving, when the brake eect given
by the essential combustion engine's motoring torque is required. A similar eect
could be achieved via ârolling in neutralâ: by opening the clutch or by shifting the
transmission into neutral the engine is decoupled from the drivetrain whenever the
driver requires a negative torque. Literature (17) (18) denes the eect of extending
the distance covered by the vehicle freewheeling, when no traction is required, by
opening the clutch automatically and removing engineâs drag torque, as âSailingâ.
We will refer to âSailingâ in the following.
As a matter of facts, the driver can experience Sailing by an increased coasting
distance, which is illustrated in 2.6: a C-segment vehicle equipped with a mid-sized
25
2 Enabling sailing for next generation stop & start
80.0%
70.0%
60.0%
50.0%
40.0%
30.0%
20.0%
10.0%
0.0%
Energy, Roadload
Energy, Engine motoring
NEDC
WLTP
Energy, Electrical load
Figura 2.5: Analysis of energy spent over cycle to propel the vehicle, motor the
engine, charge the electrical system.
Vehicle speed
diesel engine can sail for a distance more than 30% in comparison to coasting in 6th
gear, starting at 120 km/h.
Rolling in N
engine motored
6th gear
4th gear
3rd gear
0
500
1000
1500
2000
2500
3000
3500
4000
Distance [m]
Figura 2.6: Coasting distance covered by a mid-sized vehicle in dierent gears and
by rolling in neutral.
Opening the clutch under such sailing conditions gives the options either to turn
o the combustion engine or to operate it at idle speed: we will refer to as Stop &
Start Sailing and as Idle Sailing, respectively. Therefore, the potential improvement
in fuel economy by adopting Sailing could be signicant. Indeed, fuel consumption
at idle can still be benecial in comparison to a minimal fuel injection that only
26
2.2 Sailing
compensates drag losses. Though, the fuel economy potential of Sailing depends
strongly on the speed prole, i.e. driving cycle. For vehicle certication purpose,
this speed prole is dened by legislation (like WLTP or NEDC emission test cycles).
This study assess the potential fuel economy improvement of Stop & Start Sailing
in comparison to available Stop Start system for upcoming homologation speed
prole.
2.2 Sailing
A possible operating concept for sailing is presented, which could be implemented
regardless of the transmission type: i.e. such approach is feasible with any automatic transmission (DCT, MTA, AT) and manual transmission with automated
clutch. Idle Sailing can already be activated by a driver today. For a vehicle with
manual transmission, the driver could simply depress the clutch pedal, or shift into
neutral gear. Shifting from âDâ to âNâ would do the same task in vehicle with
automatic transmission. S&S Sailing, then, could be implemented and operated in
a similar manner like a Start& Stop function. Though, such approach would require
a signicant eort to the driver and change in his driving style.
Indeed, an automatic control would activate sailing according to the torque requested by the driver by pressing either the accelerator or the brake pedal. The
torque request, then, according to the vehicle speed, corresponds to a specic acceleration/deceleration. Figura2.7 compares the deceleration performed by the vehicle
in neutral and in gear. By activating sailing automatically, vehicle will decelerate as
in neutral (red line) when driver will release pedals and will be able of controlling
the vehicle deceleration by means of the brake pedal, having the driveline disengaged. If the deceleration requested via brake pedal will be lower than what could
be achieved by coasting in gear, then vehicle will exit sailing and perform DFCO.
Such activation of sailing would be benecial as for small accelerator pedal pressure
the injected fuel does not produce any positive torque at the engine crank shaft,
rather compensates the engine drag torque. Sailing would, indeed, avoid such fuel
consumption.
2.2.1 Simulation tool
Sailing control was implemented in a 1D, time based, lumped parameter dynamic
simulation of the vehicle system. Longitudinal motion of the vehicle and rotation
of the engine, transmission and driveline are modeled. The GM proprietary vehicle
energy model has been an eective tool for studying the fuel economy of various
vehicle congurations, due in large part to its modular swappable design. The simulation tool is developed in the Matlab / Simulink / Stateow environment. This
27
Speed [km/h]
2 Enabling sailing for next generation stop & start
Coasting
in N
SAILING
Coasting in Gear
Time [s]
Figura 2.7: Sailing opportunity on vehicle
model uses Simulink and custom libraries to dene the behaviour of the powertrain,
driver and vehicle. The model is organized into subsystems that include a driver, a
vehicle model and a powertrain model. The powertrain is further broken into component models that mirror physical components such as the engine, transmission,
accessories and batteries. Supervisory control functions are modelled using a combination of Simulink and Stateow. Experienced users can add, modify and improve
functions in the model. The ability to modify the model is especially important
for control and calibration optimizations. The vehicle driver model is a closed-loop
model of a human driver. The driver model controls the simulated vehicle speed to
follow the desired trace according to the mission, within a specied error tolerance.
The driver model uses acceleration pedal position and vehicle braking command
as control inputs to the vehicle and powertrain model. The simulation tool can
28
2.2 Sailing
perform computation within minutes for WLTP fuel economy cycle. The inclusion
of the Sailing model into UM is done following the modular construction of the
simulation tool so that the control strategy can be used to study other platform
combinations as well. Engine usage, energy distributions and fuel economy gures
are shown based on simulation results.
2.2.2 Simulation results
All the simulation work has been conducted on C-segment vehicle equipped with a
mid-sized turbocharged diesel engine and manual transmission.
Although fuel economy drivers might achieve signicant fuel economy drivers
might alternate light acceleration at sailing phases, such function might not be
tolerated by average drivers. The CO2 reduction potential of Sailing depends on
driving cycle. Since the vehicle speed prole during Sailing operation depends on
many vehicle parameters, e.g. tires, aerodynamics of car body, vehicle mass and
external factors like road grade, any predened speed prole with a limited speed
tolerance band inherently limits the benet of Sailing. Though, the CO2 reduction
benet of Sailing can be foreseen by evaluating the key parameters of cycles and
closely analyze speed prole.
In section âUpcoming regulatory frameworkâ, the average deceleration over cycle
reported lower values for WLTP than NEDC. In facts, while the NEDC consists of
many steep deceleration phases which can be perfectly achieved in DFCO, WLTP
shows frequent smooth decelerations which could not be achieved if coasting in gear
and would perfectly t with Sailing.
In section ??, the average deceleration over cycle reported lower values for WLTP
than NEDC. In facts, while the NEDC consists of many steep deceleration phases
which can be perfectly achieved in DFCO, WLTP shows frequent smooth decelerations which could not be achieved if coasting in gear and would perfectly t with
Sailing.
During driving, engine power is mainly used to compensate vehicle road load
(i.e. aerodynamic drag, rolling resistance and inertia) and to motor the engine.
Even though there are frequent smooth deceleration on WLTP, a positive engine
power is detected, which corresponds to the engine motoring power. In Figura2.8
sailing events can be easily detected when the engine brake power is between the
road load power and engine motoring power.
Figura2.9, then, reports the opportunity for sailing events over the whole sailing
in comparison with DFCO phases, when the driver is requested to press the brake
pedal.
29
2 Enabling sailing for next generation stop & start
0
Speed, Vehicle (km/h)
600
650
Power, Engine motoring
700
Power, Roadload F0F1F2 total
750
800
850
Power, Engine brake
900
950
1000
Time (sec)
Figura 2.8: Power request over mid-segment of WLTP
140
Motoring
Vehicle speed [km/h]
120
Sailing
Brake
100
80
60
40
20
0
0
200
400
600
800
1000
1200
1400
1600
1800
Figura 2.9: Sailing events over WLTP
2.2.3 CO2 gures
The fuel economy benet of Stop & Start varies according to vehicle mission and,
though, to the homologation cycle. Moving from the NEDC to WLTP, the time
spent at idle is reduced by half. Therefore conventional S&S system will also be
reduced by half, moving from the NEDC to the WLTP. Further enhancement, as
Extended Stop & Start, which could be implemented by a dual-solenoid stater for a
quicker engine restart, shows a positive fuel economy benet on NEDC, while it loses
much of its potential when moving to WLTP. Therefore, in order to maintain the
fuel benet of Stop & Start, engine should be stopped in other phases over WLTP.
30
2.2 Sailing
Sailing seems the right t for the upcoming regulation, enhancing the benet of S&S
by an additional 3.5%, as per Figura??. Such feature could be easily implemented
on automated transmissions, it could be implemented on manual transmission by
introducing electrically actuated clutch system.
Figura 2.10: Fuel Economy scenario of Stop & Start technology over NEDC and
WLTP
Idle sailing has not been further evaluated in this study as its fuel economy
potential over WTLP is not signicant. Tabella2.2 reports fuel economy gures in
comparison to an S&S vehicle
S&S Sailing
Idle Sailing
WLTP
−3.5%
−0.3%
Tabella 2.2: Fuel economy gures of idle sailing vs. S&S sailing
The change in the homologation procedures transition from NEDC to WLTP
will impact signicantly the fuel economy potential of todayâs Stop & Start, losing
half of its benet. Therefore, in order to maintain its favorable cost/benet ratio
such technology must extend its usage at higher engine speed. Sailing, though,
represents the next logical step in developing Stop & Start technology, eliminating
the combustion engine's drag torque.
Moreover, as such feature reports even higher fuel economy improvements than
on cycle, by increasing the distance covered without fuel injection, it might be
31
2 Enabling sailing for next generation stop & start
helpful to close the gap between publicized and real fuel consumption of vehicles,
which corresponds today to a major complain by customers.
2.3 System Requirements
Although S&S Sailing requires integration eorts on all vehicleâs subsystems, idle
sailing does not provide signicant fuel economy benet to be used on cycle. Strategies activating sailing automatically and stopping the engine without any operation
required to the driver will be essential for its introduction, in order to gain fuel
saving if operating it on cycle. Turning o the combustion engine during coasting
conditions increases the number of engine starts over vehicle lifetime, impacting
starter system, vehicle power net, steering, bracking, trasmission and HMI.
Though, Stop & Start technology aects primarily starter motors. Today they
are typically designed for Start&Stop applications for up to 300.000 engine crank
over vehicle lifetime. This number can double for Start&Stop Sailing applications.
Therefore, a new solution to crank the engine must be applied, as performing engine
start via clutch closure. We will refer to `Clutch Assisted Start' in the following.
The automation of transmission as MTA, DCT and by introducing a clutch-by-wire
system for manual transmission vehicles, paves the way in introducing such new
operations. As a matter of facts, the number of load cycles for the starter motor
can be reduced signicantly if the engine is revved up via clutch. An appropriate
clutch control strategy allows to comfortably rev up the combustion engine above
approximately 50km/h. Nevertheless a signicant increase of the starter motorâs
load cycles will remain.
Moreover, Short engine start times are of utmost importance for Start&Stop Sailing applications. They are especially challenging if an engine restart is requested
already during engine run-down, i.e. during so-called Change-of-Mind (CoM)1 situations. Clutch Assisted Start reduces the tip-in delay in such phases, by tamping
up the engine faster.
The following chapters focuses on the development clutch assisted start by applying a new methodology based on modeling and simulation.
1 Change-of-Mind
(CoM) situations occur, if the engine is about to be turned o, when all of
a sudden an engine start request occurs because the driver wishes to accelerate. This is more
common for automatic transmissions, when the driver steps on the brake pedal (triggering an
engine stop) and releases the brake pedal shortly after (triggering an engine restart). Typically,
engine start times are longer at such situations since current pinion gear starter motors cannot
engage at higher engine speeds, meaning that the starter motor can be actuated only after the
engine has completely stopped rotating. This application specic lag time of up to more than 1
second aects the driver more for Free-Wheeling applications, since vehicle response during driving
is more essential than during standstill.
32
Capitolo 3
Modeling of the vehicle driveline
Starting the engine through clutch engagement consists in spending a certain amount
of the traveling vehicle kinetic energy to spin up the engine. Clutch assisted start,
though, will cause the vehicle to decelerate and it will introduce a jerky behavior
in the drive line leading to discomfort for passengers: an appropriate clutch closing
strategy can mitigate such eect.
Hence, a torsional model of the driveline allow to investigate the impact on
drivability for dierent clutch engagement in dierent operating conditions in order
to fulll drivability requirements.
A full driveline model, from engine to wheel, has been developed in AMESim1 aims to analyze the dynamic behavior of the powertrain component.
In this chapter the modeling in AMESim of the driveline and of its components will be explained. The vehicle architecture considered refers to a conventional
front wheel drive vehicle equipped with manual transmission. The overall system
(Figura3.1) is composed of the following main elements:
1 AMESim
stands for Advanced Modeling Environment for performing Simulations of engineering systems. It is based on an intuitive graphical interface in which the system is displayed
throughout the simulation process. Using AMESim you build sketches of engineering systems
by adding symbols or icons to a drawing area. When the sketch is complete, a simulation of the
system proceeds in the following stages:
ˆ Mathematical descriptions of components are associated with the icons;
ˆ The features of the components are set;
ˆ A simulation run is initiated;
ˆ Graphs are plotted to interpret the system behavior.
33
3 Modeling of the vehicle driveline
ˆ
Engine : a 4 cylinder light-duty passenger car diesel engine, modeled in order
to compute torque resulting on crankshaft from in-cylinder gas pressure wave
and contrasting frictions.
ˆ
Starter motor : electric starter meshing with the ywheel ring, torque vs. speed
command.
ˆ
Dual Mass Flywheel (DMF) : ywheel designed as a dynamic torsional absorver,
composed by two masses, supported by a springdamping system, reducing
engine pulsation transmitted to gearbox.
ˆ
Clutch : actuated by an ono control, transmitting the portion of torque
according to his position.
ˆ
Gearbox : a manual transmission with non-innite stiness of the shafts and
ideal eciency (η = 1).
ˆ
Dierential : ideal bevel mechanism with xed ratio driving a standard dierential (50% − 50% torque distribution).
ˆ
Axel shafts : with non-innite stiness.
ˆ
Wheels and tires : wheels are modeled as rotating inertias and tires are described by torsional stiness and damping ideally without slip with the ground.
ˆ
Vehicle : considered as a translating mass and resistant force, calculated by
coast-down parameters.
3.1 Engine starting system
The starting system consists mainly of a starter motor which meshes with the ywheel and provides enough torque for the engine to win initial frictions and inertias,
reaching the ring condition Figura3.2.
The complete system model is the assembly of the components involved, considered as sub-models:
ˆ Starter motor
ˆ Flywheel
ˆ Engine
The AMESim modeling of the starting system is presented in Figura3.3
34
3.1 Engine starting system
Figura 3.1: The AMESim model of the powertrain and driveline
3.1.1 Starter motor
The torque command is achieved from the starter performance table. This table is
available in the starter datasheet and provides the values of torque delivered by the
35
3 Modeling of the vehicle driveline
Figura 3.2: Starting system schematic
motor as a function of the shaft rotary speed. The starter characteristic is plotted
in Figura3.5.
Thus the control system for the starter is a closed loop control in which the actual
velocity of the shaft is read and the torque command is given by interpolating the
performance table data. To simulate the duration of the starter action, a 0-1 (OFFON) time based signal is multiplied to the torque command. The control system is
presented in Figura3.6 and the ow of information is schematized.
The starter motor torque is delivered to the pinion, modeled by its inertia and
a gear (pitch radius available in the datasheet), which meshes with the teeth of
the ywheel ring. In order to simulate the solenoid action, which pushes the gear
to mesh, a synchronizer coupled with an idle gear simulates the engagement and
disengagement. It is directly controlled by the 0 − 1 starter motor command. A
viscous friction coecient was considered for approximating losses in bearings and
other components inside the starter. To improve the model, contact losses between
the teeth and backlash eect could be taken into account, considering non ideal
contact stiness, damping and clearance.
3.1.2 Dual Mass Flywheel
It consists of a primary mass and a secondary mass connected to each other via
a spring/damper system and supported by a deep groove ball bearing so they can
36
3.1 Engine starting system
DMF
Engine
Starter
Figura 3.3: Engine and starter motor modeling in AMESim
rotate against each other. The primary mass with starter ring gear is driven by the
engine and tightly bolted to the crankshaft.
The DMF is modeled (Figura3.8) considering its main components:
ˆ The primary mass (on the engine side) that is rigidly connected to the
crankshaft. It is considered in the overall engine inertia;
ˆ The secondary mass (on the transmission side) that is the element on
which the clutch disc engages, coupling the transmission with the engine. It
is considered as a single body with the attached pressure plate of the clutch.
ˆ The elastic elements between the two masses are a set of springs and
dampers that determine the torsional stiness of the DMF and therefore its
vibrational behavior.
The ring gear, that meshes with the starter pinion, is added into the primary
side and has the aim to deliver the torque from the starter to the crankshaft with
the correct gear ratio, determined by the ywheel working pitch radius. The synchronizer of the starter motor engages and disengages the pinion from the ring gear.
37
3 Modeling of the vehicle driveline
Control system
Motor rotational parts
inertia at pinion
Starter motor
Pinion gear
Synchronizer
Figura 3.4: Starter motor AMESim modeling
Hysteretic components of elastic dampers and internal frictions are also considered
and modeled as rotary frictions that dissipate energy. This kind of ywheel, especially in diesel engines, allows to dump down transmitted vibration to the driveline
(3.9) and improves gear change quality.
The advantage in terms of uctuation reduction achievable from the Dual-Mass
Flywheel can be observed through a comparison between the angular velocity of the
crankshaft (linked to the primary mass) and the resulting angular velocity of the
secondary mass of the ywheel. The plot in Figura3.10 shows this comparison when
the engine is in idle condition. The reduction of uctuations is remarkable, leading
to noise and vibration reduction and therefore better comfort.
Moreover, by operating as a dynamic torsional absorber, the system moves the
resonance speed range at very low values, ensuring excellent damping of engine
vibration even at idle speed. In Figura3.11 it is shown a comparison of the frequency domain, related to engine velocity, of conventional powertrain with torsional
absorber clutch disk and powertrain with DMF.
On the other hand this results in worst performance during engine cranking
38
3.1 Engine starting system
1
0.9
Torque normalized T/To
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Motor speed normalized w/wo
Figura 3.5: Starter motor characteristic curve: Torque vs. Speed
phase, as DMF shows a peak of frequency at low crankshaft speed. When the
engine is started the DMF vibration speed range is crossed. For frequent engine
starts, required for example by Start/Stop system, such event can lead to DMF
failure over time. The DMF model used for this work has a two-stage spring and
damper system. During the rst stage load is applied only on the softer springs,
while in the higher torque ranges, load is exerted by the springs with higher stiness
(second stage). The characteristic of the system is shown in (Figura3.12). Because
of internal frictions and hysteresis of damping elements, the characteristic shows a
loop in the response of the system.
Then, in order to verify AMESim model of DMF, it must replicate characteristic
torque curve. The AMESim model is tested locking the secondary mass of the
DMF while an increasing and decreasing torque is applied on the primary side.
The resulting characteristic plot of the torque vs. relative angular displacement is
presented in Figura3.13. The torque is normalized by a nominal maximum torque
T 0. The result validates the model.
In order to nd the vibrational modes of the DMF 2 during its two operating
2 AMESim
main tools for linear analysis are fast Fourier transform (FTT), Eigenvalues calculation for dierent conditions of the system, modal shapes analysis, plot of Bode, Nichols, Nyquist
diagrams and of root loci.
39
3 Modeling of the vehicle driveline
Time based
signal:
1=starter ON
0=starter OFF
Torque command [Nm]
rpm]
Motor speed [rpm]
Starter
performance
table
Figura 3.6: Starter motor control system
stages the case in which it is not coupled with the driveline is considered rst. To
induce oscillations between the two sides of the DMF, a rising and decreasing torque
with a ramp function is applied on the primary side (engine side) so that to bring the
DMF in both operational stages (Figura3.14). In Figura?? it is shown the response
to the ramp function. The variation of the velocity between the two sides indicates
the main mode of vibration of the DMF.
From the analysis, the vibration frequencies of the DMF, when working in the
rst and second stage, are detectable. The frequency value when the ywheel is
subjected to an increasing torque is slightly higher than with a decreasing torque.
This is mainly due to damping components subjected to hysteresis and frictions of
sliding parts inside the DMF (as it can be seen from the DMF characteristic chart
in Figura3.13)
40
3.1 Engine starting system
Figura 3.7: Schematic representation of Dual Mass Flywheel
Figura 3.8: AMESim model of DMF.
The Bode diagram gives a more detailed description of the frequency response of
the DMF system. Setting the input torque to the primary side as a control variable,
41
3 Modeling of the vehicle driveline
Figura 3.9: DMF uctuation reduction schematic.
Idle
Figura 3.10: Comparison between crankshaft velocity and secondary mass velocity.
42
3.1 Engine starting system
Figura 3.11: Frequency domain of DMF compared to a torsional absorber clutch
disk.
the diagrams of gain and phase variation with frequency are plotted for the output
torque to the secondary mass of DMF. For both operating stages the Bode diagrams
are plotted in Figura3.16, Figura3.17.
The maximum amplication values occurs of course at the frequencies of vibration previously seen in Figura3.15. Once the response in the frequency domain is
achieved, it can be dened also the vibrational behavior of the DMF with the engine
speed. As described in the introduction part of this chapter, the main uctuation
frequency of the crankshaft speed can be computed as:
fn =
(n[rpm])
60 · 2
(3.1)
Therefore the engine speed n expressed in [rpm] that introduces these frequencies
can be easily obtained:
n[rpm] =
43
fn
· 60
2
(3.2)
3 Modeling of the vehicle driveline
a)a
b)
Figura 3.12: a)3D schematic of two-stage spring/damper system; b)Characteristic
torque curve of the DMF
Second
stage
First
stage
Figura 3.13: Simulated characteristic torque curve of DMF
44
3.1 Engine starting system
Figura 3.14: Torque applied to DMF with an increasing and decreasing ramp
function
In real operating condition, the DMF is uncoupled with the transmission part
during engine start and idling. When idling, engine torque is quite small and overcomes only dynamic frictions and inertias. Instead engine start is a critical condition
since static friction could be very high at low ambient temperatures (even 70/80
Nm) and also the torque delivered during the rst rings, for accelerating the engine inertias, is quite high. However in normal condition the torque will hardly
be high enough to bring DMF working in the second stage. Hence considering the
DMF operating in the rst stage, from the Bode diagram, the magnitude gain can
be plotted as a function of engine speed n (rpm) (Figura3.18).
The maximum amplication occurs at an engine velocity around 300 rpm. For
such engine velocity the DMF goes into its resonance eld and is subjected to a
dangerous situation. In normal operating condition, this velocity is reached only at
engine cranking. In fact for the actual dual mass ywheels the main disadvantage
is their fragility during engine start. In Figura3.19 the velocities of the two sides of
the DMF during the engine starting phase are plotted. At around 300 − 400 rpm
the oscillations of the secondary mass becomes remarkable and the system goes out
of phase.
45
3 Modeling of the vehicle driveline
Second stage
First stage
First stage
Figura 3.15: Response of DMF to the torque applied: velocity dierence of the two
masses
3.1.3 Engine
The engine considers a 4-cylinders light-duty passenger car diesel engine. Main engine data and conguration were collected through experimental testing at engine
dyno test bench: more in details, in-cylinder pressure waves and friction data were
acquired experimentally. Since the aim of this work is to perform a torsional behavior of the starting system, the AMESim model used to simulate the engine is
a 2D crankshaft-piston model with inertias and frictions eects. It computes the
motion of a 4 pistons engine due to the kinematics and the dynamics of the engine
rotating masses and calculates the resulting torque on the crankshaft as an eect of
inputted in-cylinder pressure. The engine model set up this way cannot be considered as a closed loop system since it requires in-cylinder pressure data to run. These
data can be acquired from experimental tests or either derived from thermodynamic
equations. This second approach allows to predict pressure values and therefore to
arrange a closed loop system, avoiding (or reducing) bench test activities. For this
reason a preliminary combustion model for predicting in-cylinder pressure for dierent engine conditions is set up. In Figure 3.20 the modeling is presented, including
the main elements:
46
3.1 Engine starting system
Figura 3.16: Bode diagram of DMF in 1st stage
Figura 3.17: Bode diagram of DMF in 2nd stage
47
3 Modeling of the vehicle driveline
20
10
Gain [dB]
0
-10
-20
-30
-40
-50
-60
0
250
500
750
1000
1250
1500
1750
2000
2250
2500
2750
3000
Equivalent engine angular velocity w [rev/min]
Figura 3.18: Resonance of DMF in 1st stage as a function of engine velocity
Figura 3.19: Velocities of DMF primary and secondary mass during engine starting
phase
48
3.1 Engine starting system
ˆ 4 piston/cylinder models for crankshaft dynamic computation;
ˆ Inertia model for rotating and reciprocating masses;
ˆ Friction model;
ˆ In-cylinder pressure data.
In-cylinder pressure
Cylinder 1
Cylinder 2
Cylinder 3
Cylinder 4
input
Inertia model
Figura 3.20: Engine modeling in AMESim.
Piston, connecting-rod and crank throw assembly The crankshaft-piston
model simulates the eect of the in-cylinder pressure on the crankshaft motion.
Considering Figura3.21 schematic, the kinematic and dynamic equations can be
computed.
The instantaneous position of the piston is a function of the crankshaft angular
displacement given by:
q
2
x = R · (cos θ − 1) − L · 1 − 1 − (f racR cot sin θ − ∆L)
(3.3)
The relative velocity between the piston and cylinder walls is calculated as a function
of the rotary velocity of the crankshaft N and the lever arm done by the connectingrod associated to the crank throw G:
49
3 Modeling of the vehicle driveline
Figura 3.21: Piston, connecting-rod and crank throw assembly schematic.
R · sin θ − ∆
R
G = R · (sin θ − 1) + q
· · cos θ
L
1 − (f racR cot sin θ − ∆L)2
(3.4)
Vrelative = G · N
(3.5)
The force due to the pressure in the chamber is:
FPi = area · (Pi − Patm )
(3.6)
The corresponding torque on the crankshaft for each piston is then:
TPi = G · FPi
(3.7)
The corresponding torque on the crankshaft for each piston is then:
TPi = G · FPi
The model requires therefore the following parameters:
50
(3.8)
3.1 Engine starting system
ˆ w, the bore in mm;
ˆ s, the stroke of the piston in mm (s = 2 · R);
ˆ L, the length of the connecting-rod in mm;
ˆ ∆, the distance between the translation axis of the gudgeon pin and the crank
rotation axis in mm;
ˆ δ , the piston angle (half of the V angle) in degree;
ˆ φ, the crank throw angle in degree;
ˆ θc0 ,the initial angular displacement of the crankshaft in degree (0o = T DC ,
180o = BDC) In-cylinder pressure is provided as input to the model.
Inertias The inertia model takes into account each inertia of the system to
simulate the dynamics of the rotating and reciprocating masses. It includes
the inertia of the primary mass of the DMF. The external torque (for example
from starter motor) is added to the torque produced by the forces acting on the
piston due to in-cylinder pressure. Thus the resulting torque is used to apply
the fundamental principle of the dynamic to compute the crankshaft acceleration. The crankshaft velocity is computed by integration of this acceleration.
The kinematic constraints generates a Coriolis torque which is the product of
the square of the crankshaft velocity (N 2 ) with an inertia depending of the
crank angle Ji (θ):
Tinertiai = Ji (θ) · N 2
(3.9)
The acceleration depends both on the resulting applied torque and on the
resulting inertia which is the sum of the crank-piston inertias depending of
the crank angle.
dN
1
= P × Tr
(3.10)
dt
Ji
The crankshaft velocity is the integral of this acceleration.
The following parameters are required:
moment of inertia of the crankshaft;
moment of inertia of the primary mass of the DMF;
moment of inertia, mass and COG (center of gravity) of connecting-rod;
mass of the piston;
mass of the crank throws and position of their COG.
51
3 Modeling of the vehicle driveline
In the model it was also added an overall inertia value taking into account
all the auxiliary components and accessories directly driven by the engine
(camshafts, pulleys, belts, water pump, etc.).
Frictions From the mechanical eciency denition for an internal combustion engine, the friction work is the portion of the total indicated work, given
by the combustion, that is dissipated in a variety of ways within the engine
and auxiliaries devices and it is not available at the crankshaft. It can be
described therefore by the following aspects:
Pumping frictions due to losses during intake and exhaust phases;
Rubbing frictions due to contact between components with relative motion;
Accessory losses to drive engine essential devices for its operation.
These friction works can be considered as a negative contribution to the system, contrasting engine torque. In general the friction forces are given by
elements that are independent of speed (dry friction) and elements that are
proportional to speed (dynamic viscous friction). Therefore a speed dependent
friction model was considered with in addition a âstictionâ eect at around 0
velocity to simulate static friction that need to be overcome to enable relative
motion of stationary objects in contact. This model is shown in Figura3.22
Friction analysis is a complex topic since they depend on several factors, such
as component design, working conditions (temperature, in-cylinder pressure, load), lubrication system, leakages, etc. Thus for determining a friction
map, several bench test activities were done. Two dierent approaches were
followed: Motoring the engine, with water and oil temperature at standard
operating condition (about 90o C ), in order to measure the torque required to
overcome engine frictions at dierent velocities. This measurement represents approximately the sum of pumping, rubbing friction and auxiliary power
losses; Measuring in-cylinder pressure while the engine was running at dierent speeds and loads (bmep) allows toRdetermine directly the net indicated
power and pumping losses, integrating pdV . The friction power is therefore given by subtracting the brake power to the indicated power (in terms of
mean eective pressure: f mep = imep − bmep). While the pumping losses are
the area of P − V (Pressure vs. Volume) diagram during intake and exhaust
stroke. The friction torques measured with the rst method includes generally all kind of losses and therefore it is useful when the in-cylinder pressures,
provided as input in the piston model, are calculated from thermodynamic
equations in which for example pumping losses are not considered (intake and
52
3.1 Engine starting system
Figura 3.22: Friction model as a function of relative speed with stiction eect.
exhaust stage are ideally performed at constant ambient pressure). Therefore,
as it will be described in the following chapter, the friction values acquired
from this approach were used when the pressure data were predicted by simulation through a combustion model. Whereas measuring in-cylinder
pressure
R
allows to determine separately the pumping frictions, through pdV over the
intake and exhaust
R stages, and the sum of rubbing frictions and auxiliaries
power, through pdV over the combustion cycle, minus load (brake torque)
applied. When experimental data of in-cylinder pressure are used for running
the engine model simulation, the pumping losses are already considered in the
resulting torque computation.
Hence only the eects of rubbing frictions and auxiliaries power losses are added to the system as a contrasting eect to engine torque. In order to validate
the engine model it was necessary to manually slightly modify the friction values in order to achieve tting between simulation and experimental results.
Moreover it was not possible to have friction measurements at engine speed
lower than 700 rpm. In fact the engine shows deep vibrational problems at low
speed regimes and it is therefore avoided to run in these conditions. Moreover
the measuring instruments are not designed to perform at low velocities. Some
assumptions were made in order to complete missing data and the nal result
is plotted in the Figura3.23.
53
3 Modeling of the vehicle driveline
1
Friction with pumping loss
Friction torque normalized Tf/Tfo
0.8
Friction without pumping loss
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
-2000
-1500
-1000
-500
0
500
1000
1500
2000
Engine velocity [rpm]
Figura 3.23: Friction map from the two methods: motoring and considering pumping
losses; work integral over the combustion cycle not considering pumping losses.
The plot shows the curves of friction torque vs. engine speed resulting from
the two methods (motoring: the sum of pumping, auxiliaries and rubbing
frictions; work integral over the combustion cycle: the sum of only auxiliaries
and rubbing frictions). The friction model control system is consequently based
on detecting the instantaneous engine velocity, interpolating the friction curve
and returning the resulting torque to the system.
In-cylinder pressure In-cylinder pressure data is an input to the pistoncrankshaft model, which provides to calculate the resulting engine torque. As
a starting point, in order to validate the model, pressure inside the 4 cylinders
was measured in bench test activities during motoring and engine start phase.
It is however clear that having a closed loop system, which can simulate and
predict in-cylinder pressure waves for dierent engine conditions, can bring
to remarkable time and cost benets. Glow plug pressure sensors are used
for in-cylinder acquisitions. Two conditions were considered for validating the
whole starting system:
54
3.1 Engine starting system
1. Motoring the engine through the starter motor without any combustion
(without fuel injection);
2. Starting the engine till it reaches the idle state. For both conditions
absolute pressure data were collected as a function of crankshaft angle
position for the four cylinders. The pressure waves were analyzed and
rearranged. They are shifted of 180o for each cylinder. The ring order
is 1 − 3 − 4 − 2.
They are presented as a function of the crankshaft angle respectively in in
Figura3.24 and Figura3.25.
Pambient
-
Figura 3.24: In-cylinder pressure during engine motoring vs. crankshaft angle.
The control system for the in-cylinder pressure consists in reading the instantaneous angular position of the crankshaft and transmitting the correspondent
pressure value to the piston system.
3.1.4 Model validation
Simulations of the whole starting system model, as described, were run. The
results were compared to experimental data in order to validate the model. In
particular during test activities the engine velocity was monitored and associated to the correspondent in-cylinder pressure data collected. This velocity
55
3 Modeling of the vehicle driveline
Pambient
-
Figura 3.25: In-cylinder pressure during engine starting phase vs. crankshaft angle.
was compared to the simulated crankshaft velocity Figura3.26. Considering
the case of cranking the engine through starter motor, contact stiness between teeth and backlash can be taken into account. These eects are plotted
in Figura3.27.
For engine starting simulation, the experimental data consists of the mean
velocity of the engine crankshaft in time. The results are compared to this
available data Figura3.28.
From the result analysis it comes out that the model simulates good enough the
engine start phase. The correspondence with the experimental data validates
hence the modeling and components parameterization. The main limit of the
model is its open loop nature. That is the simulations are linked to specic
cases with dedicated data collection (as for in-cylinder pressure). Therefore it
cannot guarantee accurate results for dierent cases and conditions without
relative data. A closed loop system can be completed if the in-cylinder pressure
values are also simulated by the model. The pressure inside the cylinder during
motoring and combustion can be predicted from thermodynamics equations
and a preliminary study for this purpose is presented.
56
3.1 Engine starting system
Cranking
velocity
Figura 3.26: Engine cranking velocity with ideal gear meshing.
57
3 Modeling of the vehicle driveline
Cranking
velocity
Figura 3.27: Engine cranking velocity with stiness and backlash between gears.
Idle
velocity
Figura 3.28: Simulation of engine velocity during starting.
58
3.2 Transmission
3.2 Transmission
The engine delivers useful power to the vehicle through the transmission components. The transmission adapts the output torque of the engine to the drive
wheels. The main components are:
Clutch;
Gear box:
* Primary shaft (input shaft)
* Gears
* Secondary shaft (output shaft)
Dierential (with nal drive ratio)
Axel shafts (half-shafts)
Wheels:
* Tires.
When the clutch is engaged, the engine is directly coupled with the transmission and can deliver torque to the gearbox input shaft, which is rigidly
connected to the clutch disk. The gearbox has the function to generate a desired transmission ratio through two meshing gears. The second gear is driving
the output shaft. A set of gear ratios could be obtained selecting dierent
gears inside the gearbox. The output shaft torque is in turn delivered to the
dierential through a bevel gear with a nal drive ratio. Finally from the
dierential the torque is delivered through the axel shafts to the wheels that
provide traction to the vehicle by tire contact with the road.
A schematic overview of the powertrain is shown in Figura3.29.
A simplied schematic model of the complete powertrain is shown in Figura3.30.
Considering the transmission ratios, the angular velocity of the powertrain
parts can be related to the wheels velocity as:
Vveh = ωw · Rw where Rw is the wheel dynamic radius
ωdif f = ωw
ωgout = ωw · iF D
ωeng = ωDM F = ωclutch = ωgin = ωw · iF D · ig
(3.11)
In steady state conditions, the power transmitted from the engine available
to the wheels is aected by losses through the transmission. Therefore an
eciency coecient (ηtr < 1) can be introduced:
59
3 Modeling of the vehicle driveline
gearbox
Figura 3.29: Schematic overview of powertrain.
Engine
DMF
Input
gear
Output
shaft
ratio
shaft
final
drive Differential Wheels
ratio
Figura 3.30: Powertrain simplied model.
60
Tire
Vehicle
3.2 Transmission
Pw = ηtr · Peng
Since the power can be expressed as
Peng = Teng · ωeng
The torque delivered from the engine available to the wheels is accordingly:
Tw = ηtr · Teng · iF D · ig
(3.12)
In Figura3.31 a the engine velocity is compared to the resulting wheels velocity
when the vehicle is traveling at 50 km/h in 4th gear. While the Figura3.31 b
shows a comparison between the engine torque and the one delivered to the
vehicle.
The transmission modeling in AMESim is shown in Figura3.32.
3.2.1 Clutch
The clutch is necessary to connect the ywheel to the gearbox shaft. Its purpose is to transmit power with little slippage. This is basically done using a
friction disk that is brought gradually in contact with the two sides through a
spring (or actuator) in order to match their dierent rotating speed progressively. The clutch system is shown in Figura3.33. The clutch disc is mounted
onto the transmission input shaft and is radially xed by a splined interface.
The clutch is normally closed, as the diaphragm spring is pre-tensioned when
assembled. The axial bearing can slide over the transmission input shaft and
push against the ngers of the diaphragm spring. The direction of the release
force is swapped through the lever joints and releases the pressure from the
clutch disc, which is then able to rotate independently from the engine.
On the engine side, the pressure plate is rigidly attached to the ywheel.
For the transmission considered in this project, it is bolted to the secondary mass of the DMF. The conventional clutches are equipped with torsional
spring/damper system to reduce chatter. But when a DMF is mounted, the
61
3 Modeling of the vehicle driveline
a)a
b)
Figura 3.31: (a) Wheels velocity and engine velocity at 50 km/h in 4th gear. (b)
Torque on wheels and engine torque in 4th gear.
Figura 3.32: Transmission modeling in AMESim.
62
3.2 Transmission
a)a
b)
Figura 3.33: Clutch system schematic.
vibrations and engine irregularities are damped by the elastic elements of the
DMF itself. This allows the use of a simpler clutch that works only by contact
friction. There are essentially three distinct modes of operation of the clutch:
Free state, where the clutch is completely disengaged and transmits no
torque through the two sides, which are independent of each other.
Lockup state, where the clutch is fully engaged and the engine is rigidly
coupled to the driveline. Therefore the two sides of the clutch have the
same velocity.
Slipping state is the intermediate condition, in which the two sides have
dierent angular velocities and the clutch is acting a gradual friction force
to match them and bring to the lockup state.
The torque transmitted through the clutch is denoted by Tc . In the free state
condition it is evidently 0. While in lockup state, or sticking condition, it is
the total engine torque that is transmitted through the clutch. The sticking
of the clutch sustains as long as the torque transmitted through the clutch
remains below the maximum transmissible torque Tcmax .
This torque is given by:
63
3 Modeling of the vehicle driveline
Tcmax = Fnmax · µ0 · Ra
Tc ≤ Tcmax
(3.13)
Where Fnmax is the maximum actuation force working on the clutch plate, µ0
the static friction coecient of the clutch surface material and Ra the eective
radius of the clutch plate. In slipping state the two sides are rotating with
dierent velocities and the system is subject to dynamic variations. Assuming
a Coulomb friction model, the torque through the clutch during slipping is
given by:
Tc = Fn · µ · Ra · (ωeng − ωclutch )
(3.14)
Where Fn is the actuation force working on the clutch plate (exerted by the
diaphragm spring or by an hydraulic or pneumatic actuator system), µ the
dynamic friction coecient of the clutch and Ra the eective radius of the
clutch plate. ωeng and ωt are respectively the engine angular velocity and
the transmission input shaft velocity. The sign of their dierence suggests the
direction of the torque. For example, when the transmission velocity is greater
than the engine velocity, the vehicle inertia is driving the engine. Since the
transmitted torque is proportional to the axial actuation force Fn , the control
can be managed acting on the actuation system.
From the simulation point of view, the Coulomb model is a static model and
it is inadequate to describe dynamic systems. The Coulomb friction model
equation, as already described, can be simplied as:
(3.15)
Tc = Tdyn · (∆ω)
where Tdyn is the dynamic friction torque. The Coulomb friction model is illustrated in Figura3.34a. Because the equation of motion for dynamic systems
is strongly non-linear with a Coulomb friction model, a tanh function can be
employed to ensure the transition through zero and limit the friction torque.
The hyperbolic tangent model equation is given by:
h
Tc = Tdyn · tanh k ·
∆ω
∆ω0
i
)
(3.16)
where k and ∆ω0 are coecients that determines how fast the tanh function
changes. Tanh model is illustrated in 3.34b and compared to the Coulomb
model.
64
3.2 Transmission
a)a
b)
Figura 3.34: (a) Coulomb friction model. (b)Tanh friction model.
Another dynamic model that can be used is the Dahl model. Dahl model is
a generalization of the ordinary Coulomb equation, that is, the steady-state
version of Dahl model is the Coulomb friction. Dahlâs dierential expression
for friction is inspired from elasto-plastic behavior of materials. It is formulated
as:
dTc
dt
h
= K · ∆ω · 1 −
Tc
(∆ω)
Tdyn
i· h
1−
Tc
(∆ω)
Tdyn
i
(3.17)
Where K is the torsional stiness coecient and is a parameter that determines the shape of the stress-strain curve. In the literature, the Dahl model
is often simplied with the exponent = 1. Then:
Tc = Fn · µ · Ra · (ωeng − ωclutch )
(3.18)
The torque transmitted through the clutch can be derived from numerical
integration.
For the purpose of this project some simplications are made:
the thermal eects are not taken into account because they have a low
dynamic and inuence most of all the friction coecient µ, which is
considered constant.
the friction torque is computed using a normalized 0 − 1 command signal
(com) as a fraction of the maximum friction torque, that is:
65
3 Modeling of the vehicle driveline
Tdyn = com · Tcmax )
(3.19)
Furthermore as a starting point a tanh model was sucient to achieve proper
results. The AMESim clutch model, used in a manual gearbox, is shown in
Figura3.35.
Figura 3.35: Clutch system schematic.
The command of the clutch is 1 (100%) when disengaged and 0% when released
(engaged). The control of the clutch will be discussed in chapter 4.
3.2.2 Gearbox, dierential and nal drive
The gearbox input shaft is connected to the friction plate of the clutch. The
output shaft is driven by the input shaft through a gear mesh and is connected
to the dierential via the nal drive. The bevel gear drives the dierential
with another transmission ratio which is xed. The drive shafts connect the
satellites of the dierential to the wheels. Since only straight line driving is
considered, they are assumed to have the same velocity. The eects on torque
and speed in the gearbox and nal drive include:
Torque multiplication and speed reduction via the gear ratios;
Torque losses due to the acceleration of rotational inertias;
Torque losses due to the friction between meshing gears.
The gear ratios of the gearbox and of the nal drive are dened as:
in
ig = ωωggout
iF D = ωωggout
w
itot = iF D · ig
(3.20)
Since the gearbox input shaft is connected to the clutch, the torque delivered
to the nal drive (dierential, axel shaft and wheels) can be expressed as:
66
3.2 Transmission
Tw = ηtr · Tc · if d · ig − Jgin · ig · iF D · ωgin − Jgout · iF D · ωgout − JDif f · ω̇w
= ηtr · Tc · if d · ig − Jgin · i2g · i2F D + Jgout · i2F D + JDif f · ω̇w
= ηtr · Tc · if d · ig − Jd · ω̇w
(3.21)
where Jd (iF D , ig ) is the lumped inertia of all rotating transmission and driveline
parts, which is a function of transmission ratios.
Jd = Jgin · i2g · i2F D + Jgout · i2F D + JDif f
(3.22)
Figura 3.36: Transmission modeling in AMESim.
3.2.3 Wheels and vehicle
The drive torque Tw is the torque at the wheels. Via tire-road interaction, this
results in longitudinal acceleration of the vehicle. The model structure assumes
the driving wheels and tires as inertial components connected to the vehicle
through a linear stiness and damping. In general the governing equation of
motion is given by:
Jw · ω̇w = Tw − Rw · Fx
(3.23)
with Jw the wheels inertia, ω̇w wheels angular acceleration, Rw the dynamic
wheel radius and Fx the tire friction force. This force is dened as:
67
3 Modeling of the vehicle driveline
Fx = F2 · µ(κ, , Fz )
(3.24)
where the tire friction coecient µ(κ, , Fz ) is a non-linear function dependent
on the longitudinal slip κ, side slip angle and tire vertical load Fz . The
most common tire friction model used in the literature is the so-called Magic
Formula or Pacejka model, which uses static maps to describe the relation
between slip and friction. Though, a detailed behavior analysis of the tire
is not required to study clutch behavior. Therefore an ideal non slipping
condition is considered.
The vehicle acceleration depends on the torque delivered from the tires. A
travelling vehicle is subjected to dierent resistive loads, such as air resistance, rolling resistance, road inclination. Therefore the equation of dynamic
equilibrium can be written as:
mv · v̇v = Fx − Fres
(3.25)
with mv the vehicle mass, v̇v the vehicle acceleration, Fx the tire friction force
and Fres the resistive force. Fres can be determined through a coast-down
test3 . It denes the typical vehicle characteristics:
Aerodynamic drag;
Tire rolling resistance;
Wheel bearing and brake drag;
Axle and transmission spin loss.
These two last elements denes essentially the transmission eciency coecient (ηtr ), which is therefore already counted in the experimental Fres . Fres
is dened by three coast-down parameters (coastdown parameters) a [N],
independent on vehicle speed, b [N/(km/h)], proportional to vehicle speed
and c [N/(km/h)2 ], proportional to speed squared. Considering also a road
inclination, the formulation of Fres is:
Fres = a + b · vv + c · vv2 + mv · g · sin θ
3 Coast-down
(3.26)
test consists in launching the vehicle at a certain speed, disengage engine from
drive-line and instantaneously recording its speed and traveled distance until vehicle stops. Then
speed values over time are interpolated via a second-grade polynomial function.
68
3.2 Transmission
with vv the vehicle velocity, mv the vehicle mass, g the acceleration due to
gravity and θ road inclination. The coast-down parameters dene the resistive
load. Vehicle coasting speed as a result of the simulation (0 road inclination)
is shown in Figura3.37.
Figura 3.37: Vehicle velocity trend for a coast-down test simulation with 0 road
inclination.
From the previous equations, the relation between torque transmitted from
tires and vehicle acceleration can be found:
Jw · ω̇w = Tw − Rw · Fx
mv · v̇v = Fx − Fres
mv · v̇v = RTww · ω̇w − Fres
(3.27)
Since v̇v = Rw · ω̇w the equation can be written as:
Jw
Tw
mv + 2 · v̇v =
− Fres
Rw
Rw
(3.28)
This means that the vehicle mass and the wheels inertia can be consider as a
lumped vehicle mass mvw (slipping of the tires is neglected), expressed as:
mvw
Jw
= mv + 2
Rw
Tw
mvw · v̇v =
− Fres
Resulting in
Rw
69
3 Modeling of the vehicle driveline
(3.29)
3.2.4 Validating modeling of transmission components
The transmission components, from clutch to wheels, do not have innite stiness
and therefore are subjected to vibrational modes. As showed in ??, the rst step
to validate the model of the drive-line is to analyze vibration modes. The modeling
of the transmission describes them by their inertia and stiness (Figura3.38). The
main damping eect is achieved from the tires.
Differential
Axels
pinions
shafts
stiffness
stiffness
Tires
stiffness
Transmission
shafts stiffness
Figura 3.38: Transmission modeling in AMESim with indicated torsional stinesses.
For such task wheels are considered as a simple torsional load without slip with
respect to the ground and the torsional stiness of the tire is supposed to be constant. Gears contact stiness is also considered to be ideal since it does not inuence
signicantly the main vibrational modes. Considering the equivalent simplied model for the transmission system, the dynamic equilibrium equation is dened by Jd
(iF D,ig ). It is the lumped inertia of all rotating transmission and driveline parts,
and is a function of transmission ratios.
Jd = Jgin · i2g · i2F D + Jgout · i2F D + JDif f + Jw
(3.30)
From the denition of Jd , the inertia of the overall system depends on the nal
drive ratio iF D and on the gear ratio ig . The nal drive ratio is constant but
the gear ratio is selected by the driver. Therefore for dierent gear selections the
lumped inertia of the system changes. Since the frequency of vibration depends on
the inertia of the system, it will result in dierent vibrational modes for dierent gear
ratio selected. Therefore the frequency analysis for the drive-line is extended to each
70
3.3 Analysis of vibrations of the drive-line
of the six gear ratios for the gearbox considered. In order to visualize the vibration
of the drive-line model, a step torque is applied to the clutch. The response of the
system is shown plotting the output torque on the wheels and the Bode diagram for
each gear ratio.
The results of the frequency analysis for the driveline can be directly read in the
plots. The vibrations induced in by the step torque are displayed and the response
of the system in frequency domain is present in the Bode diagrams. Since in the
case considered the engine is not coupled with the driveline, the irregularities of its
motion doesnât aect the system.
3.3 Analysis of vibrations of the drive-line
Frequency analysis is the major point in the evaluation of the impact of powertrain
components on drivability. Every system modeled by means of inertial components
with stiness properties, is subjected to oscillatory modes. In the case of power
transmission systems using rotating shafts and couplings it is a matter of torsional
vibration, that is an angular vibration of a shaft along its axes of rotation. The
actual engine crankshaft rotation is characterized by an oscillatory motion caused
by compression and expansion strokes of each cylinder and by inertia of rotating and
reciprocating masses. Because no material can be innitely sti, these alternating
torques applied at some distance on a shaft cause twisting vibration about the
axis of rotation. These vibrations can cause attached components to fail when
the frequency of vibration matches the torsional resonant frequency. Considering a
translational spring-mass oscillators system, the general equation of motion is the
following dierential equation:
dθ
d2 θ
+ C + k = T (t)
(3.31)
dt
dt
Where θ is the angle of deection, I moment of inertia, C rotational damping, k
I
torsional stiness. In general the frequency of vibration is very near to the natural
resonant frequency of the system, that is:
1
ωn
=
fn =
2π
2π
r
k
I
(3.32)
In general mechanical resonance is the tendency of a mechanical system to respond at greater amplitude when the frequency of its oscillations matches the system's natural frequency of vibration (its resonant frequency) than it does at other
frequencies. Regarding vehicle powertrain, the designing must ensure that mechanical resonant frequencies of the components do not match driving vibrational frequencies of the resulting engine oscillating motion. For an internal combustion engine,
71
3 Modeling of the vehicle driveline
Figura 3.39: Vibrational modes of the transmission for gear 1st to 3rd .
vibration caused by the inertia of the alternating masses is less relevant at low engine
velocity with respect to the oscillation caused by gas pressure variation. Since the
project is focused on engine start phase, low velocities are considered and the main
72
3.3 Analysis of vibrations of the drive-line
Figura 3.40: Vibrational modes of the transmission for gear 4th to 6th .
contribution for vibration is due to in-cylinder gas pressure during compression and
expansion stroke.
For a 4 cylinders 4 stroke engine, at each revolution of the crankshaft (every
73
3 Modeling of the vehicle driveline
360o ), there are 2 compression/expansion stages accomplished by 2 consecutive pi-
stons in their ring order. Approximating the oscillatory motion as sinusoidal, the
main frequency involved can be calculated considering the engine speed as follows:
fn =
n
·2
60
(3.33)
where n is the engine speed in revolution per minute.
Actually the oscillatory motion of the crankshaft can be described by a Fourier
series, considering the sum of a set of simple oscillating functions. Applying a fast
Fourier transform (FFT) to the engine velocity in time it is easy to show the main
frequencies composing it. In Figura3.41 it is presented the oscillating velocity of
the crankshaft (and secondary mass of the DMF) and the resulting fast Fourier
transform diagram.
Idle
velocity
Figura 3.41: Fast Fourier transform of engine velocity at idle.
From the FFT curve it is clear that the frequency of highest amplitude is located
exactly at idle speed. Therefore as the frequencies of the engine at speed range of
operation are known, the vibrational frequencies of the driveline components are
designed not to match those of the crankshaft motion.
Friction has a damping eect, slowing the motion of the system. Damping reduces the amplitude of oscillations and it is linearly related to the velocity of the
74
3.3 Analysis of vibrations of the drive-line
oscillations. For a vehicle drive-line the main damping eect is achieved from tires.
It is very important to nd out the resonant frequencies of the drive-line in dierent
operating conditions. A frequency analysis of the drive-line focuses on evaluating
the vibrational modes of the system.
The oscillations of the drive-line is evaluated with clutch system engaged, with
engine either providing torque to the wheels or motored by the vehicle inertia.
For the aim of this project crankshaft stiness is not considered, as the main
components contributing to drive-line vibrations are the DMF, transmission shafts,
axle shafts and tires.
In ?? and ?? vibrational modes of engine and transmission are deployed with
disengaged clutch; indeed in this section, the complete powertrain system is analyzed. The case in which the clutch is engaged is the most signicant one for the aim
of this project, since it highlights the inuence of engine and clutch irregularities on
the dynamics of the vehicle. This lumped inertia of the system is the total one, in
which the engine is added, and of course dependent on the transmission ratio and
it becomes higher with high gear ratios. For each gear selected the inertia of the
system changes and therefore the vibrational mode. Due to the DMF behavior, the
system becomes more complex for frequency analysis. Three operating condition
must be considered:
ˆ DMF is not stimulated by enough alternating torque to bring a variation of
velocity between the two masses. In this case the DMF acts exactly as a
conventional single mass ywheel and it doesnât aect the vibrational behavior
of the powertrain. In this condition the free response of the system can be
obtained and its main vibrational modes.
ˆ DMF is working in the rst stage in which only the springs with low stiness
are operating. Therefore the introduction of this stiness brings to changes in
vibrational modes.
ˆ DMF is working in the second stage since higher torques are exchanged between
the two sides of the ywheel and springs with higher stiness are involved.
Therefore per each gear, these three cases of DMF operation should be considered for a complete frequency analysis. In order to visualize the vibration of the
powertrain, a step of torque is applied from the engine. By plotting the output
torque on the wheels and the Bode diagram for each gear ratio and DMF condition,
the response of the system can be evaluated.
The resonant frequencies result to be higher when the system presents high values
of stiness (DMF in second stage) and low lumped inertia (high gears), and lower
with low stiness (DMF in rst stage) and increasing lamped inertia (low gears).
As shown in Figura3.42, Figura3.43, Figura3.44, it can be observed that in every
working condition, there are two resonant frequencies that are always present. These
75
3 Modeling of the vehicle driveline
Figura 3.42: Vibrational modes of the vehicle drive-line for gear 1st to 2nd .
frequencies can be found also in the transmission vibrational analysis. Actually their
appearance is mainly due to wheels and tire stiness and can be modied using tires
with dierent properties. It is useful to know when the engine irregularities bring
76
3.3 Analysis of vibrations of the drive-line
Figura 3.43: Vibrational modes of the vehicle drive-line for gear 3rd to 4th .
the system to work in these resonant modes. As it was already mentioned, if the
system undergoes into the resonant elds for a certain amount of time, it can bring
to uncontrolled noises and also into failure of some components. When engine speed
77
3 Modeling of the vehicle driveline
Figura 3.44: Vibrational modes of the vehicle drive-line for gear 5th to 6th .
is close to idle, the power-train is in a critical condition only when high gears (5th
and 6th ) are engaged and high torque is required (DMF in second stage). Though,
this condition happens rarely in real vehicle maneuvers. Actually vibration noises
should appear because of tire stiness. When the engine is running at around 900
78
3.3 Analysis of vibrations of the drive-line
rpm the system vibrations tend to be amplied. But this noise is quite well damped
by the tire itself. The plot in Figura3.45 shows an example. When the vehicle
is travelling at 35km/h and the 4th gear is engaged, amplied oscillation can be
observed. Instead if travelling at 45km/h with the same gear, the oscillations are
lower. The resulting vibration on the vehicle acceleration is amplied for a frequency
value that matches the tires resonant frequency with respect to other frequencies.
Vehicle acceleration at 35 km/h
Vehicle acceleration at 45 km/h
Figura 3.45: Vibration of vehicle acceleration caused by matching tires resonant
frequencies.
79
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement