NT OF

ME EN
RT
A

ER
DEP

GY
UNIT

ICA
ER
ED

ST A
AT E S OF
 NOTICE

The submitted manuscript has been offered by an employee of the Midwest Research Institute (MRI),
a contractor of the US Government under Contract No. DE-AC36-99GO10337. Accordingly, the US Government and
MRI retain a nonexclusive royalty-free license to publish or reproduce the published form of this contribution, or allow
others to do so, for US Government purposes.

This report was prepared as an account of work sponsored by an agency of the United States
government. Neither the United States government nor any agency thereof, nor any of their employees, makes any
warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or use-
fulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe pri-
vately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trade-
mark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favor-
ing by the United States government or any agency thereof. The views and opinions of authors expressed herein do
not necessarily state or reflect those of the United States government or any agency thereof.

Available electronically at http://www.doe.gov/bridge

Available for a processing fee to U.S. Department of Energy

and its contractors, in paper, from:
U.S. Department of Energy
Office of Scientific and Technical Information
P.O. Box 62
Oak Ridge, TN 37831-0062
phone: 865.576.8401
fax: 865.576.5728
email: [email protected]

Available for sale to the public, in paper, from:

U.S. Department of Commerce
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
phone: 800.553.6847
fax: 703.605.6900
email: [email protected]
online ordering: http://www.ntis.gov/ordering.htm

Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste
 Page
Opening A ddress
2000 ADVISOR Users Conference: Opening Address. . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Keith Wipke, National Renewable Energy Laboratory

Par tnering with the Aut o Industry

The Rapid Development of an Electric Vehicle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Andreas Vlahinos, Advanced Engineering Solutions, LLC
David Rush, Teamwork’s Inc.
Co-Simulation of ADVISOR and Saber:
A Solution for Total Vehicle Energy Management Simulation
Presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
John MacBain, Delphi Automotive Systems
Comparison of Fuel Efficiencies and Fuel Flexibility of
Small Automotive Vehicles
Presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
John Reuyl and Robert Apter, NEVCOR Inc.

Cosimulation: Par tnering with the Softw are Industry I:

Optimization and Thermal Modeling
Implementing Optimization in ADVISOR Using VisualDOC
Presentation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
John Garcelon and Valdimir Balabanov,
Vanderplaats Research & Development, Inc.

Detailed Vehicle Thermal Systems Modeling in ADVISOR

Through Integration with FLOWMASTER2
Presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Rory Lewis and Jason Burke, Flowmaster USA, Inc.
DOE Perspectiv es on ADVISOR
ADVISOR Users Conference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .170
Bob Kost, U.S. Department of Energy

F orw ar d-Looking Simulations Coupled with ADVISOR

Simulation and Validation of Hybrid Electric Vehicles
using PSAT and PSAT-PRO
Presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .188
Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .214
Aymeric Rousseau and Maxime Pasquier, Argonne National Laboratory

The Reverse Engineering of a Diesel Engine:

A Unified Systems Approach Using ADVISOR
Presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .224
Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .249
George Delagrammatikas and Dennis Assanis, University of Michigan

New Concepts from Universities

A Hybrid-Propulsion Powertrain with Planetary Gear Set:

Simulation Results and a Design Approach
Presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .263
Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .278
Marco Santoro, Dresden University of Technology
Leone Martellucci, University of Rome

Use of ADVISOR for simulation of a Hybrid Electric Vehicle with a

Stirling Engine as the Auxiliary Power Unit
Presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .287
Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .312
Luis Figueroa and Owen R. Fauvel, University of Calgary
Graham T. Reader, University of Windsor

GWU Hybrid Electric Sport Utility Vehicle Design for

FutureTruck 2000 Competition
Presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .326
Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .337
Mohd-Syaifuddin Mohd, Zeki Gokce, Vahid Motevalli, and Kartik Bulusu
George Washington University
Validation, Vehicle Development, and Applications
Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus
Presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344
Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
Jean Bavard and Atef Gayed, Alstom Transport

Hybrid Buses - The Benefits of Matching to Real Routes

Presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
Mike Kellaway and Alan Ponsford, Newbus Technology Limited

ADVISOR Modeling of a Fuel Cell Hybrid Electric Vehicle . . . . . . . . . . . . . . . . . . . . . . 395

Mike Ogburn and Doug Nelson, Virginia Tech

Cosimulation: Par tnering with the Softw are Industry II:

Geometric and Mec hanical Modeling
ADVISOR and solid Modeler Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
Mark Henault, ESS West, Inc.

The Integration of ADAMS/Car with ADVISOR,

Realization of the "Functional Digital Vehicle”
Presentation (Joshi) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
Paper (Spagnuolo). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
Sandeep Joshi and Tony Spagnuolo, Mechanical Dynamics Inc.

Post ers
A Logic-Based, Performance-Driven Electric Vehicle
Software Design Tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
David G. Alexander and Donald M. Blackketter, University of Idaho

Co-simulation with ADVISOR and Wave: Advanced Engine Modeling . . . . . . . . . . . . 503

Trevor Blohm and Clive Hughes, Ricardo Inc.

Analysis of PNGV Inverter Power and Current Requirements

Using ADVISOR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504
Gerald W. Davis, Rockwell Automation

Development and Use of a Regenerative Braking Model in ADVISOR . . . . . . . . . . . . 508

George Delagrammatikas and Dennis Assanis, University of Michigan

A Systematic Way of Choosing Driveline Configuration and

Sizing Components in Hybrid Vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 509
Jonas Hellgren, Chalmers University of Technology
2000 ADVISOR Users Conference
Opening Address

Keith Wipke
Senior Engineer, Vehicle Systems Analysis Task Leader,
National Renewable Energy Laboratory

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

Outline
• NREL/DOE goals of conference
• Historical Perspective
• Demographics of ADVISOR Users
• Sneak-preview of ADVISOR 3.0

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

NREL/DOE Goals of ADVISOR Conference
• Bring together users from around the world to…
• Share information, models, techniques,
experiences
• Foster a sense of community, elevate the visibility
of the virtual community that already exists
• Celebrate successes and identify opportunities for
improvements
• Get feedback on future direction for ADVISOR and
the Digital Functional Vehicle process

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

 Historical Perspective:
Evolution of ADVISOR in 6 Years

t o ols
3 .0 els sim.
of s 2. 0 I O R od er c e
t I R GU eb I S d M ot h r en
m en gin G U O
IS ed n w ADV rove with nfe
p be . 0 V c o p ng o
lo
ve SOR
1 AD han sed I m k i er s
C
e
D VI OR En lea in
VI
S e L s t Us
AD A D R 1

9 94 9 95 9 96 9 97 9 98 9 99 0 00
1 1 1 1 1 1 2

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

ADVISOR User Demographics

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

ADVISOR Being Used Globally
November 1998: ~130 users

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

ADVISOR Being Used Globally
January 1999: ~330 users

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

ADVISOR Being Used Globally
March 1999: ~500 users

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

ADVISOR Being Used Globally
August 1999: ~800 users

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

ADVISOR Being Used Globally
August 2000: >2000 users

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

ADVISOR Downloads by Continent

38 25 10
231

28

Africa

Asia

Australia

Europe

No. America

So. America
459
No Country
922

Legend includes total number of

users on 7 continents or 1713
users.
As of 6/9/00
NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS
ADVISOR Downloads by Country
25 25 22
27
33
39
42 United States of America
United Kingdom
Korea
51
Italy
Canada
51 Germany
France
China
Japan
63
Sweden
Netherlands
No Country
72 841 India
Australia

90

Legend included countries with 20 or more users

111 Total number of countries using ADVISOR = 70
Total users in all countries = 1713

As of 6/7/00
NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS
ADVISOR Downloads
by Type of Organization
64 40

Industry
University
Govt/Labs
Other

597

1012

As of 6/7/00
NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS
ADVISOR Downloads by Industry
5 5
5 Ford Motor Company
5
5 47 DaimlerChrysler Corporation
6 General Motors
7 Visteon
Delphi Automotive Systems
6 Volvo Car Corporation
FIAT Research Center, Italy
6 Hyundai Motor Company
AVL
7 Parametric Technologies Corporation
Siemens Automotive Systems
8 29 AlliedSignal
Honda
Renault
9 Technologies M4 Inc.
Epyx Corporation
Hino Motors, Ltd.
12 Nissan Motor Company
Self-Employed
25 TNO Automotive
13
Legend includes industry with 5 or more users
13 Total number of industries using ADVISOR = 577
22 Total users in all industries = 1012
15

As of 8/18/00
NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS
ADVISOR Downloads by Universities
5
5 5 27
5
5 Ohio State University
6 University of Michigan
University of Maryland
6 University of Bath
George Washington University
6 26 University of California - Davis
Texas Tech University
6 Georgia Institute of Technology
Cornell University
7 University of Tennessee
Texas A&M University
Hanyang University
7 Pennsylvania State University
San Diego State University
22 Seoul National University
7 West Virginia University
Anna University
7 University of Colorado
University of Sheffield
Virginia Tech
7
16 Cranfield university
Institute for Advanced Engineering
8 MIT
University of Kent at Canterbury
University of Leeds
9 15
9
11 15
14 Legend includes universities with 5 or more users
Total number of universities using ADVISOR = 277
Total users in all universities = 597
As of 6/7/00
NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS
ADVISOR Downloads by State
11 11
13 12
15
17
14
23 211 Michigan
California
New York
26 Ohio
Texas
Colorado
27 Maryland
Massachusetts
Virginia
28 Indiana
Pennsylvania
Georgia
Tennessee
33 District of Columbia
Wisconsin
Illinois
Minnesota
36 Florida
110 Washington

36
Legend included states with 11 or more users
38 Total number of states using ADVISOR = 45
41 37 Total users in all states = 841

As of 6/7/00
NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS
ADVSIOR Downloads
by Government Agencies/Labs
2
2
8
2 Argonne National Lab
National Renewable Energy Laboratory
2 Oak Ridge National Lab
Natural Resources Canada
Politecnico di Torino
2 Southwest Research Institute
TASC
USDOE
2 California Air Resources Board
8 EPA
Los Alamos National Laboratory

3 Netherlands Energy Research Foundation

Southern California Edison
USDOT

3
Legend includes government agencies/labs
with 2 or more users
Total number of government agencies/labs
3 7 using ADVISOR = 64
Total users in all agencies/labs = 27
3
4
As of 6/7/00
NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS
Multiple Versions Downloaded as one
Indicator of “Active” Users
• ~20% appear to be “active” with ADVISOR
Number of ADVISOR Versions Downloaded (8/17/00)

2500
2078
2000

1500
# Users

1000

412
500
109
21 5
0
1 or more 2 or more 3 or more 4 or more 5 or more
versions versions versions versions versions
Series1 2078 412 109 21 5
# Versions

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

Multiple Versions Downloaded as one
Indicator of “Active” Users
• ~20% appear to be “active” with ADVISOR
Number of ADVISOR Versions Downloaded (8/17/00)

450 412
400
350
300
# Users

250
200
150 109
100
50 21 5
0
2 or more 3 or more 4 or more 5 or more
versions versions versions versions
Series1 412 109 21 5
# Versions

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

Preview of ADVISOR 3.0

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

Vehicle Input Screen

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

Simulation Setup Screen

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

Cycle Results Screen

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

Future Direction: Digital Functional Vehicle

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

Special Thanks...

• You, Our Users, for making it possible

• Valerie Johnson, Barbara Ferris (NREL) for
organizing conference
• ADVISOR Development Team

Clockwise: Ken Kelly, Sam

Sprik, Keith Wipke, Tony
Markel, Valerie Johnson, Aaron
Brooker, Terry Hendricks
(not pictured: Michael O’Keefe)

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

Special Thanks...

• Our ADVISOR Users Conference 2000 Sponsors:

NREL, CENTER FOR TRANSPORTATION TECHNOLOGIES AND SYSTEMS

 The Rapid Development of an Electric Vehicle
Andreas Vlahinos, Advanced Engineering Solutions, LLC
David Rush,Teamwork’s Inc.

Co-Simulation of ADVISOR and Saber:

A Solution for Total Vehicle Energy Management Simulation
Presentation
Paper
John MacBain, Delphi Automotive Systems

Comparison of Fuel Efficiencies and

Fuel Flexibility of Small Automotive Vehicles
Presentation
Paper
John Reuyl and Robert Apter, NEVCOR Inc.
 Advisor User Conference
August 2000

The Rapid Development of

an Electric Vehicle
Dr. Andreas Vlahinos, Principle
& David Rush, President & CEO
Advanced Engineering Solutions
& Teamwork’s Inc.

Your Virtual Resource for Rapid New Product Development

Emerging Electric Personal Mobility

From Electric Bikes

To Electric City Vehicles

 Emerging Electric Personal Mobility
To everything in between
 Emerging Electric Personal Mobility
To everything in between
So in such a tough competitive
emerging market ....

The one that is . . . .

First to market . . . .

with a quality product . . . .

at low cost.

So . . . . how do we get there?

 Industry Benefits
Brand name recognition
– Early pricing premiums
– Greater market share
– Reputation as innovator
– Greater profitability

Quality Product
– Product quality is no longer an option, It is a way of life
– Quality is a measure of satisfaction of the customer’s
requirements & expectations
– Quality is dynamic (changing customer expectations)

Customer Loyalty
– 70 percent of customers bail because of the look/feel/smell/taste
of doing business with a company

Shareholder Value
– mission statement “ … to make a profit…”
 The Process
The activities and procedures (and any sub
process) that supports the vehicle
conceptualization, it’s definition, design and
development including :-
– Targets process.
– Quality Operating System.
– Review & decision processes.
– Development process & procedures.
– Sign-off process, etc.
Classic Systems Engineering Process

Define customer Reqmts

Vehicle Reqmts
Develop &
System Reqmts Optimize Design
to meet Reqmts
Component Reqmts

Set, cascade
& balance
targets Verify & Confirm
design meets
requirements
A Successful Process requires….
• Customer requirements drive the process
• Well defined process is key to rapid product
development
• One process – no mavericks
• People must understand and use the process
• Compatible tools
• Integrated methods & techniques
• Web based communication
“Knowledge has become the key economic
resource and the dominant, if not the only,
source of competitive advantage”
Peter Drucker

• Virtual prototypes
• Tool selection for Integration
• Directional indications on time is better than
absolute results to late
• Avoid multiple masters - single data base
 The People
“ In the digital age, as we move into
the quicker and quicker exchanges of
information . . . and re-inventions of
the world at work, our organizations
and our careers in action will become
more and more closely aligned with
the jazz ensemble . . . we will find
ourselves improvising with greater
and greater confidence and fearing
less and less the imaginative power of
the individual committed to enriching
the whole.”

Stanley Crouch, Forbes

 The People
It is in the team environment that all other tools are
most effectively used.

– Team structure.
– Integrate product attributes & system design & release
responsibilities.
– Minimize the number of teams required.
– Minimize static data with real time communications.
– Provide central program data access.
 The People
People Skills provide . . .

• Knowledge of tools and methods.

• Skilled in application within the emerging
integrated tool sets.
• Capability to collaborate through distributed
development processes.
• Competency to develop new skills in an ever
changing technical environment.
 The People – Team Structure
Product Director

Vehicle Verification Team

Business Team
Proj. Mgmt & Timing Team Prototype & Test Planning Team
Marketing Team
Homologation Team
Business Structure Team Program Management
Team
Water Management Team
Reusability Team
Technical Environmental Cubing
Sourcing Team

Vehicle Weight Team

Package
Program Systems Support
Informations Systems Vehicle Integration
Team (VIT)
Vehicle Configuration
CAD/CAE/CAM Technology
Elec/ElecArchitecture
Concept Appearance Development
Climate Control Architecture
Appearance Color Harmony

WORK TEAMS

Structure/Topology Team Interior Team Exterior Team Cockpit Team Vehicle Dynamics Team Powerpack Team
 The People – Team Structure
C o n c e p t Ap p ea ra n c e D ev elo p m en t
C lim a te C o ntrol A rc h ite c tu re
A p p ea ra n c e C o lor H arm on y

W O R K TE AM S

S tru c ture/T op o log y T e a m Inte rior T e am E xte rio r T e am C o c k p it T e am V eh ic le D yn a m ic s T ea m P ow erp ac k T ea m

A ttrib u te s C h a s s is N V H P ow e rtrain N V H

B o dy N V H Inte rio r A ir T h e rm a l/A e ro H u m an F a c to rs

R ide & H a nd lin g PED

S afe ty
S te erin g

S ec u rity/A nti-T h e ft
B ra k ing

M o d u le T e a m s
B o d y S tru c ture In terio r T rim F rt E n d S ys In t. F rt S us p P ow e r U n it

F ixed G las s R e ar S e ats R r E n d S ys In t. R r S u s p T ra ns m is s io n

2n d U n it B o dy H e a dline r E xt. O rn . T ub e B u nd le D rive lin e & A xle

B od y C los u re F rt S ea ts & R e s tra in ts T ires & W h ee ls

E ng . C om p t. D re s s
(inc . E xh au s t)
F ram e C a rp e t-N V H -W irin g C oc k p it F ue l S to rag e

R e a r R e s tra ints C o n s ole

C o m p o n e n t D e s ig n G ro u p s (C D G ) N a tu ra l W o rk G ro u p
C o m po n en t CAD M a nu fac tu ring CAE A ttrib ute A s s em b ly T es t
R e le as e E n gin e er D es ig n ers P ro c e s s E n gin e er E n gin e er D e v. E n gin ee r P ro c es s E n g ine e r E ng in ee r
The People & Vehicle Integration

Structure Topology

Power Propulsion

Vehicle Dynamics

Interior

Exterior

Cockpit
 Multi-Functional Attribute Balancing
Manufacturing Shared Structural
Supplier Guidelines Knowledge Performance
t>2, R>1 (Local Buckling, tf/tw) Targets (m,Kt,Kb,ω
ω)

Economic Analysis
Cost (material, Beam Analysis
fabrication, assembly) Behavioral Design of
Modeling Experiments
Styling
Packaging
Optimum Beam Sections Topology
Optimum Connection Shape Optimization
New Approach
CAD Design
FEA / Optimization Meet Final Chassis
Body in White
Body in White Target CAD Design
“ Product quality requires managerial, technological
and statistical concepts throughout all the major
functions of the organization …”
Josheph M. Juran

Variation (thickness, properties, surface finish,

loads, etc.) is … THE ENEMY

DOE, Six Sigma, Statistical FEA, Behavioral

Modeling … THE DEFENCE
Behavioral Modeling of a Typical Section

•All Sections have the same moment of Inertia

•Find the one that minimizes the cross sectional
area (Min Weight) and meet all the manufacturing
and stability requirements
•Not a dimension driven CAD model
•Requirement driven design (Ireq)
Topology Optimization Short Section
( finding the best use of material )
Topology Optimization Long Section
Suspension Characteristic Targets
mass, inertia, Global Geometry info

Handling, Ride and

Suspension type & Steer Characteristics
Hard Point Location

Mechanism Motion
Dynamic Loading
Space Claim Envelop

CAD Design FEA / Optimization Meet Final

Suspension Suspension Target Suspension
?
Solutions for optimal cost without sacrificing ride & handling
 Energenix Center

Co-Simulation of ADVISOR and Saber

A Solution for Total Vehicle Energy Management Simulation

John MacBain
August 24, 2000

Co-Simulation of
Page 1 ADVISOR and Saber
 Agenda

◆ Overview
◆ Evolution of Automotive Architectures 1
– Traditional ala 2000 - sizing batteries and generators
– Traditional with increased electrical power budget
◆ Co-Simulation Concept
◆ Evolution of Automotive Architectures 2
– Dual Voltage Architecture
– Series Hybrid
– Parallel Hybrid
◆ DOE Contract for Co-Simulation of ADVISOR and
Saber
– Status Report

Co-Simulation of
Page 2 ADVISOR and Saber
 Overview

◆ Electrical architecture simulation has traditionally

been independent from the propulsion system of
the vehicle
◆ Increasing electrical power budgets in traditional
vehicles (EVA, EPS, catalytic converter heating,
etc.) make consistent solution of the propulsion
and electrical systems necessary for accurate
results (mpg, sizing of electrical components,
macro power flow, etc.)
◆ Hybrid architectures effectively marry the
electrical and propulsion system, making them
inseparable from a computational standpoint

Co-Simulation of
Page 3 ADVISOR and Saber
 Traditional 14V Architecture

Drive
IC Engine Transmission Rear Differential Tires Cycle

Generator Ignition System

Other Electrical
14V Battery Loads

Figure 1: Traditional Vehicle Architecture

ICE rpm determined largely by drive cycle

Generator load largely does not effect ICE rpm
Drive cycle related loads are largely the ignition
system
Co-Simulation of
Page 4 ADVISOR and Saber
 Traditional 14V Architecture
◆ Sizing of batteries and generators - a key
simulation activity at the macro power level

– Select challenging temperatures and a variety of drive cycles

– Select minimum battery capability based upon specifications
– Select the electrical loads for each drive cycle

– Simulate the propulsion system for the each drive cycle

– Convert ICE rpm profile to a generator rpm profile
– Simulate performance of electrical system for each drive cycle
– Check adequacy of generator to maintain battery charge
– Adjust size of battery or generator accordingly and repeat the
simulation cycle

– Non-interactive analyses function adequately

Co-Simulation of
Page 5 ADVISOR and Saber
 Traditional 14V Architecture
Increased Electrical Power Budget
◆ Three driving factors in future vehicle design
– Projected generator peak power requirements
» 1,800 watts in 1990
» 10,000 watts in 2020
– Tightening CAFÉ standards
– Tightening emission standards

◆ Implications:
– Events in the propulsion system and electrical system have
increasing impact on the other
– Simulations of the total power system of the vehicle, electrical
and propulsion, must be more interconnected as we design to
meet stringent requirements requiring system solutions on a
broader scope

Co-Simulation of
Page 6 ADVISOR and Saber
 A Possible Solution
◆ Potentially Ideal solution - model electrical system in
MatLab/Simulink as a part of ADVISOR

◆ Challenges with the ideal solution

– Saber and other packages already are developed and focused on
the solution of the electrical system
– Many automotive OEMs are committed to Saber for electrical
system analysis
– Many component models have already been developed in Saber
and not in MatLab
– Saber imports Pspice models

◆ Thus, it makes sense to connect existing

specialized tools rather than re-inventing the wheel

Co-Simulation of
Page 7 ADVISOR and Saber
 Co-Simulation Concept
Exchange Parameters at Each Time Gate

Electrical System Propagation

Saber

Propulsion System Propagation

time = T ADVISOR time = T + delta T

Independent Propagation
During Each Time Step

Co-Simulation of
Page 8 ADVISOR and Saber
 Co-Simulation Concept
Traditional Vehicle Architecture
Potential Parameters to Pass:
ICE instantaneous rpm
Generator instantaneous required shaft torque

Exchange Parameters at Each Time Gate

Electrical System Propagation

Saber
Propulsion System Propagation

time = T ADVISOR time = T + delta T

Independent Propagation
During Each Time Step

Co-Simulation of
Page 9 ADVISOR and Saber
 Dual Voltage Architectures
IC Engine Transmission Differential Tires Drive Cycle

ADVISOR
Co-Simulation Link
42V Generator Significant 42V Loads 14V Electrical
Related to Propulsion Loads
System

42V Battery
14V Battery

Other 42V Electrical Loads

Saber DC to DC
Converter

Co-Simulation of
Page 10 ADVISOR and Saber
 Series Hybrid Architecture

Drive Motor Transmission Differential Tires Drive Cycle

Either

Significant Electric Loads

ADVISOR

Saber
Battery Generator IC Engine

Co-Simulation of
Page 11 ADVISOR and Saber
 Parallel Hybrid Architecture

Drive Motor / Transmission Differential Tires Drive Cycle

Generator

Either
Significant Electric Loads

Saber
IC Engine Battery

ADVISOR

Co-Simulation of
Page 12 ADVISOR and Saber
 Initial DOE Contract Activities

◆ Kickoff on June 8, 2000

◆ Demonstrate restart of Saber with parameter

alterations
– Commands issued within Saber

◆ Start drafting Saber AIM script for co-simulation

as a shell around Saber
– Commands in effect coming from outside Saber

Co-Simulation of
Page 13 ADVISOR and Saber
 Restart Demonstration of Saber
1st .5 seconds

Show graphs 3 and 5

Expanded View of Startup

Co-Simulation of
Page 14 ADVISOR and Saber
 Restart Demonstration of Saber
1st .5 Seconds
Field Voltage at 14V

Restart Runs to 1 second

Field Voltage drops to 10V
Co-Simulation of
Page 15 ADVISOR and Saber
 Ideal Co-Simulation Strategy

ADVISOR
MatLab /Simulink Saber

Parameter Exchange
by Direct Communication to
Command Lines

Co-Simulation of
Page 16 ADVISOR and Saber
 More Realistic Ideal Co-Simulation Strategy

AIM Script Shell

ADVISOR
MatLab /Simulink
Saber

Parameter Exchange
with AIM Script Shell

Co-Simulation of
Page 17 ADVISOR and Saber
 1st Prototype Strategy for Co-Simulation

Data Set #1 AIM Script Shell

ADVISOR
MatLab /Simulink Saber

Data Set #2
Prototype AIM Script
Parameter Exchange validated in August by
via Data Sets Joe Conover.
Co-Simulation of
Page 18 ADVISOR and Saber
 Conclusions

◆ Future evaluation of electrical systems and

propulsions systems will require acknowledgement
of the codependency of the two systems
◆ It makes sense to build upon existing specialized
software systems
– Co-Simulation provides a viable method to achieve this by
establishing communication between existing software packages

◆ The DOE contract for co-simulation of ADVISOR

and Saber has been started and progress has been
made.

Co-Simulation of
Page 19 ADVISOR and Saber
 Co-Simulation of ADVISOR and Saber - A Solution for Total
Vehicle Energy Management Simulation

John A. MacBain, Ph.D.

Delphi Automotive Systems, Energenix Center

ABSTRACT INTRODUCTION

Electrical analysis for vehicles with traditional 14V This paper will explore the largely independent
electrical architectures has often been limited to requirements on electrical system analysis today. Once
engineering functionality simulation for the components this baseline has been established, the paper will move
and systems. This level of analysis ignores the impact through a progression of vehicle architectures from
of the electrical system on the fuel efficiency and traditional architectures with higher electrical power
emissions of the full vehicle, a much larger system. budgets to future hybrid concepts with the perspective of
Even for traditional architecture vehicles that are requirements for electrical system analysis. This
prevalent today, the computational problems of increased electrical power budget comes from the shift
predicting propulsion system performance and electrical of loads from mechanical powered loads to the electrical
system performance are coupled. As the millennium system (EPS, EVA, etc.), the additional of new loads
changes, series and parallel hybrid vehicle concepts are (catalytic converter heaters), and hybrid vehicle
becoming a marketplace reality. These hybrid concepts architectures with electric traction motors. The
accentuate the coupling of the propulsion and electrical propulsion system and the electrical system become
systems making it less viable to simulate either ever more interdependent, moving electrical system
propulsion or electrical systems independently. This analysis from a largely independent activity today to an
paper explores the various goals of electrical system embedded activity with total vehicle propulsion system
analysis and how some electrical analyses require performance. The discussion establishes the argument
acknowledgement of the coupling of the electrical and for co-simulation between traditional electrical analysis
propulsion systems thus creating a requirement for software and traditional propulsion system analysis
simultaneous or co-simulation of the propulsion and software. The paper will end with a discussion
electrical systems. concerning the specific plans of the contract between
Delphi Automotive Systems and the National Renewable
This paper will lay out the plans for developing specific Energy Laboratory (Department of Energy funding) to
co-simulation technologies between NREL’s ADVISOR make co-simulation between ADVISOR and Saber a real
and Saber. This work represents the content of a and available option in future versions of ADVISOR.
contract between Delphi Automotive Systems and the
Department of Energy that establishes a partnership
between the National Renewable Energy Laboratory and
Delphi Automotive Systems. The contract is entitled TRADITIONAL 14V VEHICLES – SIZING
“Modeling and Simulation Development for Advanced ELECTRICAL COMPONENTS
Energy Management and Propulsions Systems.”
A traditional 14V system typical of most passenger car
The final product of the contract will be a co-simulation vehicles on the road today has a single voltage electrical
capability tying Saber (for the electrical architecture system which drives all the electrical loads on the
simulations) and ADVISOR (now for the purely vehicle. A schematic for traditional vehicles is displayed
propulsion system simulations) to produce total vehicle in Figure 1. The energy input to the electrical system is
energy management system simulations for traditional, through the shaft of the generator which is turned by the
series, and parallel hybrid vehicles. The work will internal combustion (IC) engine. The electrical
provide templates to accommodate both single and dual architecture consists of a generator (including voltage
voltage electrical architectures. The co-simulation regulator and rectifier), a storage battery, and a series of
capability will be available from NREL to the global electrical loads. This section will discuss this electrical
automotive engineering community much as ADVISOR architecture with the perspective of performing various
is available today with template interfaces for electrical analyses often faced in vehicle design.
constructing and simulating vehicle energy management
systems. A variety of electrical system problems are regularly
solved without acknowledging the larger vehicle
systems, and often this is well justified. Vehicle
designers always face the challenge of adequately sizing
the generator and the battery to satisfy the electrical
power demands of the vehicle. More detailed models of When a vehicle is sized for a generator and battery, the
the components are required for analyses which would engineer must select several typical driving cycles. The
indicate interference of signals between components engineer also selects several extreme cases such as
and/or the impact of anomalous voltage events triggered being stuck in rush-hour traffic. Typical load scenarios
by the switching of significant electrical loads. are then selected which would cause a strain for the
Challenging problems also exist in the choice of the electrical system. For instance, vehicles must continue
proper cables for the power and signal distribution to perform in the hottest desert environments in Death
system. One must carefully choose the cable gauge, Valley as well as harsh blizzard conditions in Alaska and
insulation class, and circuit fusing methodology (level Canada. Anticipated driver load choices (AC in summer,
and fuse/circuit breaker style) to insure proper and safe etc.) and drive cycle required loads (brake lights, turn
performance of the electrical system with adequate signals, etc.) must be included.
current and voltage being made available to the various
loads in the vehicle. The balance of this section of the The typical analysis methodology proceeds with an
paper will elaborate more on the proper sizing of independent propulsion system analysis showing the
generators and batteries since more awareness of these crankshaft rpm required for the vehicle to meet the
issues will provide the basis for later topics in the paper. requirements of the driving cycle. The crankshaft rpm is
All of these problems are readily addressed as converted to generator shaft rpm reflecting either a direct
independent electrical system analyses. These analysis mechanical drive system or a belt drive. This generator
technologies have already been created with a high level rpm is the sole input to the electrical system analyses.
of sophistication within Delphi Automotive Systems and The electrical system is analyzed for battery and
other companies. generator adequacy for the various load scenarios
coupled with the drive cycles and ignition system
Electrical loads can be categorized as loads which run electrical loading consistent with the drive cycle.
continuously during vehicle operation, loads which run
for prolonged periods, loads which typically run for The engineer adopts a margin of safety for the
briefer periods, and key off loads. Loads which run component specifications and the electrical system
continuously would include the ignition system, the typically works quite well. Analysis has played an
electric fuel pump, electric fuel injection systems, and important role in this engineering methodology for many
engine management computers / controls / sensors. years, and it has proven to be quite successful. For
Loads which run for prolonged periods would include the these purposes, independent electrical system analysis
windshield wipers, headlights, taillights, side marker is quite adequate.
lights, instrument panel lamps, entertainment systems,
engine cooling fans, and the HVAC blower. Loads with As Figure 1 demonstrates, the electrical and propulsion
briefer duty cycles include power windows, power door systems are indeed largely autonomous except for the
locks, power seats, horn, starting motor, brake lights, mechanical connection between the IC engine and the
backup lights, heated rear window, and the power generator where the engine speed is largely determined
antenna. These lists have been presented for by the drive cycle. The luxury of this relative autonomy
clarification purposes and are not exhaustive. is facing extinction. In Figure 1 as with all other figures,
the arrows represent flows of information in ADVISOR
and between ADVISOR and the electrical simulation.

Drive
IC Engine Transmission Rear Differential Tires Cycle

Generator Ignition System

Other Electrical
14V Battery Loads

Figure 1: Traditional Vehicle Architecture

TRADITIONAL 14V VEHICLES – VEHICLE Today, though, packages tend to specialize in
EFFICIENCY ANALYSIS MOVING FORWARD mechanical systems or electrical systems. Two good
examples are the ADVISORâ software from NREL that
Let us raise the bar for electrical analyses. In addition to analyzes propulsion system performance and Saberâ
designing a well-performing electrical system, we need that is naturally cast towards solving electrical system
to ascertain the impact of the electrical system on the problems from its roots in the electronics business.
overall vehicle fuel efficiency and emissions. Such From a theoretical perspective, one has a series of
challenges have become reality in recent years with coupled models that must be solved simultaneously.
tightening CAFÉ standards, tightening EPA emission However, some models are being solved by software
standards, and the perpetually increasing electrical package A and the others are being solved by software
power budgets on vehicles. This electrical power budget package B. The answer is co-simulation. Co-simulation
has increased from a generator peak power of 1.8kW in is the closest you can come to simultaneous simulation
1990 to a projected 10kW in 2020. This increasing while maintaining independent solvers.
demand for electrical power is one of the primary drivers
for consideration of dual voltage or high voltage systems
for future vehicles; the dual voltage systems permit
delivering increased power at reduced current and,
therefore, reduced resistive losses and improved CO-SIMULATION OF ANALYSIS PACKAGES
efficiency. Can the engineer adequately solve this
problem by analyzing the electrical and propulsion Simulation models exist in several distinct classes.
systems separately? Let us explore the factors involved Many processes on a vehicle can be described by
in this challenge. equations, either differential or algebraic. Other
processes are more easily modeled empirically with
The primary coupling of the propulsion and electrical lookup tables of measured or pre-computed
systems is the transfer of energy from the engine performance. The solution of a simulation problem with
crankshaft to the generator. The energy being supplied time as the independent variable involves stepping a
to the generator shaft is the only energy input to the solution forward in time which satisfies all of the models
electrical system. But, how realistic is the computed rpm and their respective interactions and interdependencies.
profile of the engine crankshaft for a given vehicle on a As an example consider solving a branched circuit
drive cycle? It is typically computed through an involving resistors, capacitors, and inductors. Kirchoff’s
independent powertrain analysis that at best has an laws permit writing the coupled equations for the voltage
average power loss to the generator determining energy drops and currents. The subsequent numerical solution
consumption. Careful electrical system analysis can comes readily from packages such as Saberâ and
compute the actual energy consumption of the electrical Pspiceâ that are designed for solving circuit problems.
system including generator losses. This mechanical Or, the differential equations themselves could be
loading of the engine can then be iterated back into the modeled in any number of other software packages such
propulsion system analyses, thus updating the engine as CSSLâ (Continuous System Simulation Language) or
energy consumption. Certainly the fuel consumption
MatLab/Simulinkâ.
profile of the engine will change, and potentially the
transmission shift profile and engine rpm profile will
The problem becomes more complex when the
change as a result. One can iterate between electrical
equations cannot be conveniently solved simultaneously
system analyses and propulsions system analyses,
by a single solver. This is the challenge faced when a
providing updated profiles of the power gain/loss
vehicle’s electrical system is modeled in Saberâ and the
between the two systems until convergence is achieved.
Only then will the propulsion system analysis yield the propulsion system in MatLab (ADVISORâ from NREL).
correct fuel economy and emissions for the drive cycle A good approximation to a simultaneous solution can be
and electrical load assumptions. achieved with a technique known as co-simulation.

Now, let us move five years into the future when the Co-simulation is a computational strategy that repeatedly
power budget of a typical car has increased employs a simple algorithm that is very straightforward
substantially. The vehicle architecture may be much the to describe. The schematic of co-simulation is depicted
same for low to mid range vehicles. But, now the in Figure 2. Co-simulation is a process where the two
electrical budget will be much more significant in the solvers move forward independently through a time step.
total power budget of the vehicle. At this point, there is This time step may be user specified or it may be
no alternative to solving the electrical and propulsion theoretically controlled depending upon the dynamics of
problems simultaneously. This would be quite the vehicle system at that point during the drive cycle.
straightforward if all the models and/or differential The time step must be sufficiently small so that
equations resided in a single piece of software with a parameters of mutual interest do not experience
single solver. significant changes. At the end of the time step, key
information must be exchanged between the packages
to update the coupling or linking of the two solutions.
With the electrical analysis updated with refreshed The coupling becomes tighter with the 42V/14V vehicle
propulsion system parameters (IC engine rpm as certain 42V loads are added. Future 42V loads will
determining generator rpm) and the propulsion system certainly include electrical valve actuation (EVA), electric
updated with refreshed electrical system parameters brakes, heated catalytic converter, and electric power
(power drain from the IC engine to drive the generator), steering. These subsystems provide further couplings
a restart capability is exercised for each software between the propulsion system and electrical systems,
package to solve for next time step. This recurring representing a shift from traditional mechanically
sequence of parallel solutions through a time step powered subsystems to electrically powered
followed by an updating of parameters of mutual subsystems. The EVA system presents a speed
significance continues until the drive cycle has been dependent electrical load. Power steering and braking
completed. are considered propulsion for this discussion since they
traditionally draw power directly from the IC engine.
Exchange Parameters at Each Time Gate With this migration from mechanical power sources to
electrical power sources, the electrical power budget of
the vehicle may climb as high as 10 kW. The net result
is that independent propulsions system analyses and
Electrical System Propagation
electrical system analyses will be less representative of
reality and the requirement for co-simulation becomes
more significant.

Propulsion System Propagation SERIES HYBRID ARCHITECTURES

time = T time = T + delta T A series hybrid architecture is a propulsion system
architecture where energy is created and stored in one
Independent Propagation
form which is then utilized by a separate propulsion unit
During Each Time Step
to create motive power. A potential hardware
implementation would have an IC engine running at a
Figure 2: Schematic for Co-Simulation
highly efficient setting to drive a generator to create and
store electricity. The main propulsion unit would be an
electrical drive system running from a battery (and
generator) power. Figure 4 depicts the architecture of
DUAL VOLTAGE VEHICLES – 42V/14V such a vehicle.

Dual voltage electrical architectures face potentially The electrical and propulsion systems appear
more complex challenges than traditional 14V significantly more coupled than for the traditional vehicle
architectures. The first difference is the obviously more because the electrical system now embodies a
complex electrical system analysis. Figure 3 displays a significant component of the propulsion system. A
possible schematic for the electrical system. As a matter typical series hybrid will require co-simulation. The
of observation, such architectures will often include two control strategy / algorithm may well derive inputs from
batteries, battery control algorithms to enhance battery both the propulsion system and the electrical system.
life and performance, DC to DC converter(s), and a For instance, the control algorithm for the electric drive
much increased electrical power budget as a proportion motor may well have inputs from an automatic
of the total vehicle power budget. transmission to facilitate smooth shift points. Or, a low

IC Engine Transmission Differential Tires Drive Cycle

42V Generator Significant 42V 14V Electrical

Loads Related to Loads
Propulsion System
42V Battery
14V Battery

Other 42V Electrical Loads

DC to DC
Converter
Figure 3: 42V/14V Vehicle Architecture
charge state might force limitations in torque production PARALLEL HYBRID ARCHITECTURES
that would influence the transmission shift points (limp
home state). In this case, the coupling of the propulsion A parallel hybrid architecture is a propulsion system
and electrical systems requires co-simulation. Other which can draw motive power individually or in
options such as regenerative braking or electric power combination from two distinct powerplants. A possible
steering would also dictate co-simulation. configuration would be an IC engine and an electric drive
motor on the same drive shaft. The electric drive motor
I would be remiss if I did not observe that under certain would provide motive power in the regimes where the IC
limiting assumptions, the particular coupling of systems engine is less efficient. That same drive motor would
in a series hybrid can amount to complete decoupling of then serve as a generator to charge the battery while the
the problems from a computational standpoint. If the IC engine is providing the motive power. This same
control of the IC engine, the generation of electrical drive motor also serves as the flywheel and starter
power, and the battery control algorithm are dependent motor. Figure 5 depicts a typical parallel hybrid
solely upon electrical system parameters, and certain schematic.
electrical loads crossing between the electrical and
propulsion sides are not present (electric brakes, electric Parallel hybrid systems provide the most complete
power steering, etc.), then the problems can be coupling of the propulsion and electrical systems. A
completely decoupled. If one views the propulsion control algorithm shifts the motive power from the IC
system as everything downstream of the electric drive engine to the electrical drive motor to situations where
motor (not including the drive motor), then the necessary both are contributing. The control algorithm governs “IC
speed and torque profile for the electric drive motor can engine off” states, controls the smooth restart, and
be determined for the vehicle to match a drive cycle. A creates a seamless system for motive power generation
subsequent and completely independent analysis of the responsibility. Co-simulation is the only viable approach
electrical system (including the electric drive motor and if true simultaneous solution is not possible.
the IC engine) can determine the viability of the control
algorithm, battery, generator, and IC engine. Please
note that co-simulation is still possible, and the results
should be very comparable. So, the potential for
decoupling would not prevent full integration of the two
simulation techniques.

Drive Motor Transmission Differential Tires Drive Cycle

Significant Electric Loads

Battery Generator IC Engine

Figure 4: Series Hybrid Architecture

Drive Motor / Transmission Differential Tires Drive Cycle

Generator

Significant Electric Loads

IC Engine Battery

Figure 5: Parallel Hybrid Architecture

ALTERNATIVE VEHICLE ARCHITECTURES CONCLUSIONS

Similar discussions could be written relative to electric All analyses to determine vehicle energy efficiency over
vehicles, fuel cell / electric series hybrid vehicles, and a drive cycle require careful analyses of the propulsion
the myriad of other nuances of hybrid concepts. The and electrical systems. The optimal situation is the
story will be the same for all. There is a deep coupling simultaneous solution of the equations governing all the
of the propulsion and electrical systems. Any attempt to processes describing the creation, conversion, storage,
analyze the efficiency of the entire vehicle relative to a and application of power. When the opportunity for
drive cycle will require simultaneous solution of all the simultaneous solution in a single solver becomes
relevant equations. When simultaneous solution is not impractical, then co-simulation involving multiple distinct
possible, co-simulation becomes a requirement. solvers becomes the option of choice. With sufficiently
small time steps, the solution resulting from co-
simulation should provide a very good approximation to
the true “simultaneous solution” results.

CO-SIMULATION PLANS – ADVISOR AND Co-simulation of ADVISOR and Saber is becoming a

SABER reality for future versions of ADVISOR through a contract
entitled “Modeling and Simulation Development for
Advanced Energy Management and Propulsions
The active contract “Modeling and Simulation
Systems” between Delphi Automotive Systems and the
Development for Advanced Energy Management and
Propulsions Systems” will create the reality of co- National Renewable Energy Laboratory (Department of
Energy funding).
simulation between ADVISOR and Saber. Through a
template interface much like that presently utilized in
ADVISOR, the future user will have the option to model CONTACT
the electrical system in Saber and co-simulate the
propulsion system and the electrical system. This will John A. MacBain holds BS degrees in physics and
make existing component models available in ADVISOR mathematics from Case Institute of Technology (1971),
such as the ISET (Institut für Solare an M.S. and Ph.D. in applied mathematics from Purdue
Energieversorgungstechnik) battery models available for University (1974), and an MSEE from the University of
improved representation of the electrical architecture. Dayton (1978). John served in the Air Force as an
Moreover, for traditional vehicles, the future ADVISOR Associate Professor of Applied Mathematics at the Air
customer will be able to represent time varying electrical Force Institute of Technology. Since that time, John has
loads on the IC engine based upon electrical load worked in industry spending eight years in advanced
switching and the interactions of the generator and seismic and electromagnetic exploration development in
battery. Templates will be provided for a user interface, the oil industry. The balance of the time has been with
and generic models will be provided much as the General Motors / Delphi Automotive with assignments
present ADVISOR has generic modules. Work on the ranging from the GM Research Laboratory to managing
contract began in June, 2000, and is scheduled to take the Low Observable program at Allison Gas Turbines.
two years for completion. John’s current work assignment is in the Energenix
Center where he has responsibilities for systems
analysis. John can be reached at on the internet at
[email protected].
 Comparison of fuel efficiencies and fuel flexibility of
small automotive vehicles

Advisor Users Conference

Costa Mesa, California

August 24, 2000

John Reuyl
Robert Apter
NEVCOR Inc.

Dr. John Reuyl Dr. Robert Apter

PO Box 18683 28 Ave. 4-Marronniers
Stanford, CA 94309 1400 Yverdon , Switzerland
Tel/fax 1 650 701 0464 Tel/fax 41 24 426 7160
[email protected] [email protected]
 Key Points

• HEVs can reduce fuel use in two ways:

– by improved fuel efficiency, and
– by recharging from the electric utility grid.
• By reducing fuel use, HEVs reduce CO2 (global greenhouse gas)
• By reducing fuel use, HEVs also reduce NOx (key ozone precursor)
• HEVs can be a source of electricity for the grid (as well as recharging
from the grid).
• Many HEV designs are likely to find profitable market niches

NEVCOR, Inc.
 Table 1 - Vehicle Fuel Efficiency (mpg)(1,2)

CV FCV SEV PEV BEV

Drive Cycle 830 kg 1030 kg 1030 kg 930 kg 1030 kg

CYC_HWFET 52.5 84.7 39.3 55.9 139.6

CYC_FUDS 41.4 75.7 37.2 53.8 122.5

CYC_NEDC 40.9 81.1 38.9 50.9 129.2

CYC_1015 36.1 58.2 39.6 52.0 143.0

(1) See Appendix A for Advisor simulation parameters

(2) See Appendix B for NREL calculation of “gasoline equivalent fuel
consumption” for FCV and BEV

NEVCOR, Inc.
 Table 2 - Fuel Efficiency of FCV, SEV, PEV and BEV
Compared to CV(1,2)
Drive Cycle FCV/CV SEV/CV PEV/CV BEV/CV

CYC_HWFE 1.6:1 0.8:1 1.1:1 2.7:1

CYC_FUDS 1.8:1 0.9:1 1.3:1 3.0:1

CYC_NEDC 2.0:1 1.0:1 1.2:1 3.2:1

CYC_1015 1.6:1 1.1:1 1.4:1 4.0:1

(1) See Appendix A for Advisor simulation parameters

(2) See Appendix B for NREL calculation of “gasoline equivalent fuel
consumption” for FCV and BEV

NEVCOR, Inc.
 Figure 1 - Comparison of Range and
Refueling/Recharging Times for CVs, EVs and HEVs

500 Level 1 Recharging, 1.4 kVA @ 120 V/12 A

500 Level 1 Recharging, 1.4 kVA @ 120 V/12 A
450
450 Level 2 Recharging (Residential), 7.7 kVA @ 240 V/32 A
Level 2 Recharging (Residential), 7.7 kVA @ 240 V/32 A
400
400
Level 2 Recharging (Commercial), 11.5 kVA @ 240 V/48 A
Level 2 Recharging (Commercial), 11.5 kVA @ 240 V/48 A
350
350
Level 3 Recharging (3-phase/480 V "fast-charge energy station"),
300 Level 3 Recharging (3-phase/480 V "fast-charge energy station"),
1/2 battery capacity in 10 minutes
miles

300 1/2 battery capacity in 10 minutes

Range,miles

250 Gasoline/Diesel Refueling (Fillup in 5 minutes)

250 Gasoline/Diesel Refueling (Fillup in 5 minutes)
EV: Battery pack = 20 kWh
Range,

EV: Battery pack = 20 kWh

200 Electrical efficiency = 4 miles/kWh
200 Electrical efficiency = 4 miles/kWh
Range = 80 miles
150 Range = 80 miles
150
CV and HEV: Fuel tank = 15 gallons
CV and HEV: Fuel tank = 15 gallons
100 Fuel efficiency = 30 mi/gal
100 Fuel efficiency = 30 mi/gal
Range = 450 miles
50 Range = 450 miles
50
0
0
0 1 2 3 4 5 6 7 8
0 1 2 3 4 5 6 7 8
Refueling or Recharging time, hours
Refueling or Recharging time, hours

NEVCOR, Inc.
 Figure 2 - Distribution of Personal Automobile Use

100%

80%

Cumulative 60%
Percent
of Personal
Automobiles 40%

20%

0%
0 30 60 90 120 150 >155
Average Daily Travel Distance per Vehicle (miles)
Source: 1990 Nationwide Personal Transportation Survey (Ref. 6)

NEVCOR, Inc.
 Figure 3 - Personal Automobile Miles Electrified

100%

80%

Vehicle 60%
Miles
Electrified HEV
(%) 40%
BOEV

20%

0%
0 20 40 60 80 100 120 140
Battery Range for BOEV and HEV (miles)
Source:1990 Nationwide Personal Transportation Survey (Ref. 6) and NEVCOR.

NEVCOR, Inc.
 Control of NOx emissions is central to the control of
regional air pollution

• From Rethinking the Ozone Problem in Urban and Regional Air

Pollution (1992 National Academy of Sciences):
• NOx is a major precursor to regional ozone and particulate matter.
• “In the presence of anthropogenic NOx and under favorable
meteorological conditions, background biogenic VOCs can contribute
to summertime ozone concentrations exceeding the NAAQS
concentrations of 120 ppb.”
• Hence, NOx control is necessary to achieve air quality standards.
• “Except in California, NOx emissions reductions have not been a major
component of most state implementation plans (SIP). Hence, efforts to
achieve national air quality standards have largely failed.”
• Even in California, standards for NOx emissions from mobile sources
have lagged actual accomplishments of the auto industry (next slide).

NEVCOR, Inc.
 California NOx standards actually lag the
accomplishments of the auto industry

0.40
Notes:
0.35
1) 1998 V-8 Mercury Grand Marquis is
0.30 a full-sized 6-passenger automobile.
2) California ARB does not require any
0.25 SULEVs until 2004.
NOx, g/mi

0.20

0.15

0.10

0.05

0.00
TLEV LEV/ULEV SULEV 1998 V-8
MERCURY
California ARB NOx Standards Adopted in 1998

NEVCOR, Inc.
 ARB still considers the BEV to be the “Gold Standard”

• ARB considers the BEV to be the “Gold Standard” for two reasons:
– 1) inherent emissions durability, and
– 2) extremely low fuel-cycle emissions in California.
• However, the ARB admits that it is quite upbeat regarding
emissions durability of catalysts; ARB observes that: ...with proper
fuel tailoring adjustments, these latest technology palladium-
rhodium designs lose virtually none of their emissions conversion
capability over more than 100,000 miles of aging (1998 LEV II
amendments).
• Regarding low fuel-cycle emissions, the ARB is right; in fact, about
75% of all NOx emissions associated with electricity used by BEVs
in the SoCAB will actually be generated outside California, and
fuel-cycle emissions now dominate total emissions (next slide).

NEVCOR, Inc.
 NOx emissions of hypothetical hybrid SUVs that would
match CA NOx of a BEV in the SoCAB in 2010

29 mi/gal Diesel Vehicle, 8.1 l/100 km Total World NOx other than US
Tailpipe NOx 0.01 g/mi.

46.4 mi/gal Gasoline Vehicle, Total US NOx except California

5.1 l/100 km
Tailpipe NOx 0.01 g/mi.

20.5 mi/gal Diesel Vehicle, Total California NOx other than

11.5 l/100 km South Coast
Tailpipe NOx 0.005 g/mi.
Vehicle Tailpipe NOx (g/mi)
32.8 mi/gal Gasoline Vehicle,
7.2 l/100 km
Tailpipe NOx 0.005 g/mi.
Total South Coast Fuel-Cycle
Total NOx for 5 mi/kWh EV, NOx
12.4 kWh/100 km
Scenario 4

13 mi/gal 98 Ford Expedition 18.1 l/100 km

(Gasoline) in Scenario 4

multiply NOx scale by

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30
0.62 to obtain g/km
NOx, g/mi.

NEVCOR, Inc.
 Key Points

• HEVs can reduce fuel use in two ways:

– by improved fuel efficiency, and
– by recharging from the electric utility grid.
• By reducing fuel use, HEVs reduce CO2 (global greenhouse gas)
• By reducing fuel use, HEVs also reduce NOx (key ozone precursor)
• HEVs can be a source of electricity for the grid (as well as recharging
from the grid).
• Many HEV designs are likely to find profitable market niches

NEVCOR, Inc.
 Comparison of Fuel Efficiencies and Fuel Flexibility of
Small Automotive Vehicles
John Reuyl and Robert Apter
NEVCOR, Inc.

ABSTRACT DIVERSITY OF THE MOTOR VEHICLE MARKET AND

THE OPPORTUNITIES FOR HEV TECHNOLOGIES
ADVISOR is used to simulate hybrid-electric vehicles
(HEV) that reduce fuel use through improved fuel Hybrid-electric vehicles (HEV) can be configured in a
economy or by recharging from the grid. “Flexible-fuel” variety of ways, and, as the HEV technology matures,
HEVs are shown to “electrify” more miles than each market sector (e.g., personal vehicles, buses,
equivalent battery-electric vehicles (BEV), yet also travel trucks) will have different configurations.
long distances like a conventional vehicle (CV) but with
improved fuel economy. The paper shows how HEVs can Even within a single sector, such as personal vehicles,
result directly in lower emissions of greenhouse gases different HEV configurations are likely to appeal to
and ozone precursors (e.g., NOx). HEVs also could serve different buyers.
as a source of electricity for the grid. The paper
concludes that no single HEV design will be best; This diversity should not come as a surprise. For
instead, many designs could find profitable market example, although today’s auto industry has been
niches. “maturing” for more than 100 years, the marketplace is
filled with a variety of offerings, from economy
INTRODUCTION transportation for less than USD 10,000 to custom
vehicles costing more than USD 100,000.
“Motor vehicles and the global environment are on a
collision course. Motor vehicles are a major source of Within this market segment for personal vehicles, there
NOx [nitrogen oxides], the key precursor emission to the are very different technologies: minivans, vans, muscle
formation of low-level ozone. As a result, most major cars, station wagons, sport utility vehicles (from 500 kg
metropolitan areas experience periods of unhealthy air micro vehicles to 3,000 kg trucks), sedans of all kinds
quality each year. On a global basis, carbon dioxide and sizes from economical 4-, 5-, and 6-passenger family
(CO2) levels increase each year, due in large measure to sedans to luxury vehicles and limousines.
the increasing use of fossil fuels, and there is increasing
consensus among researchers that climate change is Even within any one very narrow niche of this market,
occurring, in part, from these anthropocentric emissions remarkable diversity exists. Consider the choices within
of CO2 and other “greenhouse” gases. There also is the the category of SUV (or minivan, or sedan). Small,
knowledge that the increasing per capita consumption of medium, large, 3-door, 4-door, 5-door; even within any
fossil fuels in the face of an increasing global population one model (e.g., 4-door sedan), engines come in 4-
is not sustainable. cylinders, 6-cylinders, 8-cylinders; with or without
supercharging; and the consumer choices continue for
“Yet the growth of modern economies has been number of valves, cams, fuel systems, transmissions,
dependent in part upon the expanding supply of energy. etc.
Of special need has been the secure supply of
petroleum-based fuels for cars, trucks and buses.” This variety shows no signs of diminishing, and the HEV
(Ref.1) technologies open the doors to even more differentiation.
As the HEV technologies mature, we can expect
This paper describes a variety of HEV technologies that manufacturers to compete for market share with HEV
can ameliorate these issues by reducing fuel use, features that are fine-tuned to individual buyers.
reducing ozone-precursor NOx emissions and increasing
the use of nighttime grid electricity. This paper will illustrate this diversity with small vehicles
as an example.
In addition, while HEVs may become the next generation
of mobility with “quantum leaps” in efficiency, they also
could be used as a source of electricity for the grid.
HEVS CAN REDUCE FUEL USE AND CO2 Note that there are additional vehicle attributes that are
not discussed in this paper, such as acceleration,
ADVISOR SIMULATIONS braking, etc., that could differ from one vehicle type to
another. The comparisons of efficiency are, however, of
Simulations with ADVISOR for fuel efficiency were increasing importance and a major element of HEV
completed for small vehicles, beginning with an 830-kg design.
baseline conventional vehicle (CV). The competing
vehicles were FCV (fuel cell), SEV (series-hybrid), PEV In Table 2 the fuel efficiency of the competing vehicles
(parallel-hybrid) and BEV (battery electric). The key (FCV, SEV, PEV and BEV) are compared to the baseline
vehicle parameters are summarized in Appendix A. Each CV.
of the vehicles was similar except that the weights were
adjusted to account for the different propulsion systems. In the sections that follow, the results in Tables 1 and 2
will be discussed and conclusions will be drawn.
The results for fuel consumption are shown in Tables 1
and 2. NREL has provided a methodology whereby the HEVS CAN REDUCE FUEL USE THROUGH
“gasoline equivalent fuel consumption” of FCVs and IMPROVED FUEL EFFICIENCY
BEVs can be compared to CVs (see appendix B).
The fuel cell HEV (FCV) that was modeled (see Table 1)
(1,2)
Table 1 - Vehicle Fuel Efficiency (mpg) offers much higher fuel efficiency than the equivalent CV.
The superiority was the greatest in the European NEDC
CV FCV SEV PEV BEV drive cycle wherein the FCV had double the fuel
economy of the CV. (see Table 2).
Drive Cycle 830 1030 1030 930 1030
kg kg kg kg kg
Such improvements in fuel efficiency result directly in the
CYC_HWFET 52.5 84.7 39.3 55.9 139.6 reduction of 1) the use of petroleum fuels and 2) the
CYC_FUDS 41.4 75.7 37.2 53.8 122.5 associated generation of CO2.
CYC_NEDC 40.9 81.1 38.9 50.9 129.2
Such improvements in fuel economy will also be shown
CYC_1015 36.1 58.2 39.6 52.0 143 to reduce ozone-precursor emissions (e.g., NOx).
(1)
See Appendix A for simulation parameters
(2)
However, FCVs face formidable technical and economic
See Appendix B for calculation of “gasoline challenges, while HEVs using combustion engines are
equivalent fuel consumption” for FCV and BEV
already in the market. These HEVs offer performance
and utility that are comparable to conventional vehicles,
and they, too, can offer substantial improvements in fuel
Table 2 - Fuel Efficiency of FCV, SEV, PEV and BEV efficiency.
(1,2)
Compared to CV
In addition, technologies developed for HEVs (motors,
Drive Cycle FCV/CV SEV/CV PEV/CV BEV/CV controllers, energy storage technologies) will also be
applicable for FCVs (and BEVs), thereby increasing their
CYC_HWFET 1.6:1 0.8:1 1.1:1 2.7:1
marketability.
CYC_FUDS 1.8:1 0.9:1 1.3:1 3.0:1
CYC_NEDC 2.0:1 1.0:1 1.2:1 3.2:1 The PEV has particular appeal because it offers
substantially improved fuel efficiency when compared to
CYC_1015 1.6:1 1.1:1 1.4:1 4.0:1 a comparable conventional vehicle, especially in the
(1)
See Appendix A for simulation parameters urban drive cycle. Furthermore, the simulations
(2)
See Appendix B for calculation of “gasoline summarized in Table 1 were done using the same
equivalent fuel consumption” for FCV and BEV conventional internal-combustion engine in the CV and
the HEVs. As will be presented in a future paper (see
Ref. 2), the engines for HEVs in the future will offer vastly
more options than will be the case for CVs. Such
The drive cycles are standards in the motor vehicle
optimization will permit further improvements in fuel
industry and are well-known to vehicle designers. The
economy for the HEVs.
US CYC_HWFET has the greatest fraction of highway
travel, the US CYC_FUDS and European CYC_NEDC
The SEV that was simulated using the standard
have a mix of urban and highway, and the Japanese
ADVISOR parameter selection shows an improvement
CYC_1015 is an urban cycle.
over the CV only in the urban drive cycle. However, in
proprietary work done at NEVCOR using ADVISOR,
designs for small SEVs have been simulated that do BEVS THAT RECHARGE FROM THE GRID REDUCE
provide much better fuel efficiency than the optimized CV FUEL USE
baseline - typically more than 20% better for the NEDC
drive cycle. The NEVCOR designs include improvements BEVs provide mobility but use grid electricity instead of
in a number of vehicle systems including energy storage fuels. Since most electricity is generated from sources
technology, engine control strategy and vehicle other than petroleum, BEVs actually displace fuel use by
optimization priorities. using grid electricity instead.

In addition, it will be shown that so-called “grid- In addition, BEVs produce zero emissions at the vehicle,
connected” SEVs can reduce fuel use by a surprisingly an attribute that gained political prominence in California
large amount. in 1990 (see Ref. 3). During the past 10 years, since the
mandate of zero-emissions vehicles (ZEV) by the
California Air Resources Board (Ref. 3), BEV
development has been pursued by virtually every major
motor vehicle manufacturer.

However, BEVs have proved to be limited in their appeal

because of their a) range limitation (typically 100-200
km), b) high cost (substantially more than an equivalent
CV with no range limitation), and c) lengthy recharge
time (see Figure 1).

Figure 1 - Comparison of Range and Refueling/Recharging Times for CVs, EVs and HEVs (from Ref. 10)

500 Level 1 Recharging, 1.4 kVA @ 120 V/12 A

500 Level 1 Recharging, 1.4 kVA @ 120 V/12 A
450
450 Level 2 Recharging (R esidential), 7.7 kVA @ 240 V/32 A
Level 2 Recharging (R esidential), 7.7 kVA @ 240 V/32 A
400
400
Level 2 Recharging (Commercial), 11.5 kVA @ 240 V/48 A
Level 2 Recharging (Commercial), 11.5 kVA @ 240 V/48 A
350
350
Level 3 Recharging (3-phas e/480 V "fast-charge energy station"),
Range, miles

Level 3 Recharging (3-phas e/480 V "fast-charge energy station"),

Range, miles

300 1/2 battery capacity in 10 minutes

300 1/2 battery capacity in 10 minutes
250 Gasoline/Diesel Refueling (F illup in 5 minutes)
250 Gasoline/Diesel Refueling (F illup in 5 minutes)
E V: Battery pack = 20 kWh
E V: Battery pack = 20 kWh
200 E lectrical efficiency = 4 miles /kWh
200 E lectrical efficiency = 4 miles /kWh
Range = 80 miles
150 Range = 80 miles
150
CV and HE V: F uel tank = 15 gallons
CV and HE V: F uel tank = 15 gallons
100 F uel efficiency = 30 mi/gal
100 F uel efficiency = 30 mi/gal
Range = 450 miles
50 Range = 450 miles
50
0
0
0 1 2 3 4 5 6 7 8
0 1 2 3 4 5 6 7 8
R efueling or R echarging time, hours
R efueling or R echarging time, hours
As a result, recent work has focused on a) longer range HEVS THAT RECHARGE FROM THE GRID ALSO
batteries (which are typically even more costly than the REDUCE FUEL USE
baseline Pb-acid technology), b) “rapid” recharge stations
(which are not as rapid as refueling and provide much Like BEVs, an HEV that can recharge overnight from the
less range than a full tank of fuel, see Figure 1) and c) electric utility also reduces the use of petroleum. If such
numerous studies to determine just how much range HEV technologies were commercialized, such vehicles
might be enough. could become a major new nighttime load on the grid,
thereby displacing petroleum use, leveling loads, and
If these issues can be satisfactorily resolved and BEV improving utility profitability. This scenario is developed in
costs can be acceptable for at least some consumers, more detail in this section.
then BEVs that be charged from the electricity grid offer
an exceptional technology for improved “gasoline An HEV with a relatively small (but high power) battery
equivalent fuel efficiency” (see Tables 1 and 2). Potential pack could travel all the short around-town trips and local
applications in the near-term include post office fleets, commutes on electricity alone. In a conventional car (or
parcel delivery, shuttle buses, and numerous specialty in an HEV when using fuels), these short trips are the
vehicle applications (e.g., lift trucks, aircraft tow vehicles, ones during which engines are cold, traffic is stop-and-
trash collection). go, mileage is poorest and emissions are greatest. In
such a “grid-connected” HEV, the fueled-engine may not
In the next section it will be shown that HEVs that can be be needed at all for the shorter trips, serving instead as a
recharged from the grid offer the same benefits supplementary source of energy for the longer trips.

Figure 2 - Distribution of Personal Automobile Use (from Ref. 4)

100%

80%

Cumulative 60%
Percent
of Personal
Automobiles 40%

20%

0%
0 30 60 90 120 150 >155
Average Daily Travel Distance per Vehicle (miles)
Source: 1990 Nationwide Personal Transportation Survey (Ref. 6)
The number of vehicles that travel short distances each improving utilization factors and utility efficiency.
day is surprisingly large. Even in the US where average
auto trips are the longest, Figure 2 (from Ref. 4) shows With its own on-board engine, such an HEV has no need
that more than 40% of all personal automobiles on the for daytime recharging or an expensive daytime
road on a typical day (roughly 40 million autos) actually recharging infrastructure. For days and even weeks at a
travel that day 20 miles or less. time, the engines in such HEVs may never be used, but
they would always be available, providing the same “full-
Such short distances are well within the feasibility of tank” feeling of security offered by the conventional
today’s EV and HEV technology using relatively small, vehicle.
low-cost battery packs.
Just as auto manufacturers offer a variety of
However, the majority of these same 40 million vehicles engine/transmission options for today’s buyer, such
will also travel much longer distances on weekends, on HEVs could be offered with a variety of battery and
vacation trips and on occasional side trips or engine options to meet tomorrow’s consumer choices.
emergencies during the work week. For example, the battery size and battery technology
could be selected based on each buyer’s likely average
The dilemma is that short-range BEVs cannot be used daily use. Buyers may choose engine size in much the
for these longer trips. Therefore, BEVs have been fitted same way as in a conventional vehicle; small engines for
with large, heavy (and expensive) battery packs in hopes the economy-minded, larger engines for those expecting
of providing enough range to satisfy at least some to haul trailers or do a lot of mountain driving.
potential users.
According to the research in Ref. 4, even HEVs with very
Yet HEVs can be optimally configured to meet both of small battery packs could electrify a very large fraction of
these very disparate duty cycles; small, low-cost battery total vehicle miles. “For any given battery range, HEVs
packs for the short trips and small engines that can could electrify, on average, more miles than BOEVs
provide the average power for unlimited long-range trips [battery-only electric vehicles] … because HEVs can be
whenever necessary. driven on all trips, and the initial miles every day will be
powered by utility electricity.” (Ref. 4, pg. 12). Figure 3
Figure 1 shows that recharging overnight for the short (next page) is reproduced from Ref. 4; it shows that,
trips would be possible from any 110V receptacle, compared to a fleet of conventional cars, a fleet of HEVs
without the infrastructure expense for a) dedicated 220V that plugged in every night and had only 30-mile battery
circuits, b) high-power chargers or c) elaborate custom packs could electrify roughly 50% of total vehicle miles
interconnect hardware and controls. Such overnight traveled.
charging would help to level utility loads, thereby

Figure 3 - Personal Automobile Miles Electrified (from Ref. 4)

100%

80%

Vehicle 60%
Miles
Electrified HEV
(%) 40%
BOEV

20%

0%
0 20 40 60 80 100 120 140
Battery Range for BOEV and HEV (miles)
Source:1990 Nationwide Personal Transportation Survey (Ref. 6) and NEVCOR.
Even if fuel efficiency when using fuels were unchanged remain.... In the presence of anthropogenic NOx and
in these “flexible-fuel” HEVs from today’s fleet average of under favorable meteorological conditions, these
27.5 mi./gal, gasoline use would be reduced by about the background biogenic VOCs can contribute to
same 50% as a fleet of HEVs with doubled fuel efficiency summertime ozone concentrations exceeding the
(to 55 mi./gal) but no ability to recharge from the grid. NAAQS concentration of 120 ppb.” (Ref. 5)
Obviously, were fuel efficiency to be improved in these
HEVs (as is expected), fuel use would be further The California Air Resources Board (CARB) also
reduced. concluded that “...because ozone precursors, such as
NOx also react in the atmosphere to form particulate
The HEV also provides the same benefits of “gasoline matter (PM), reductions in NOx will be crucial to meet
equivalent fuel consumption” as the BEV when using existing state and federal PM10 standards, as well as the
grid electricity. Note from Table 1 that both the SEV and new federal standards for fine particulate matter
the BEV weigh the same 1030 kg. Thus, the SEV using (PM2.5).” (CARB, Ref. 8)
grid electricity would have roughly the same “gasoline
equivalent fuel efficiency” as the BEV when compared to However, “Except in California, NOx emission reductions
a similar CV traveling the same route. have not previously been a major component of most
[State Implementation Plans] SIPs.” (Ref. 5) As a result,
Note that the “grid-connected, flexible-fuel” HEV provides “Despite the major regulatory and pollution-control
the user with mobility even when fuel shortages occur. programs of the past 20 years, efforts to attain the
While longer weekend trips may be curtailed, the National Ambient Air Quality Standard of ozone largely
essential daily work-related trips can be undertaken have failed.” (Ref. 5)
using grid electricity.
Based upon the conclusions of the NRC, the authors
Such HEVs could vastly increase the market share for have focused their emissions work in the remainder of
EVs beyond that for just the BEV. HEVs that can this paper on NOx emissions.
recharge at night from the grid could be the most
practical way in the near-term of displacing transportation VEHICLE NOX EMISSIONS CAN BE REDUCED TO
fuels with utility electricity. NEGLIGIBLE LEVELS

HEVS CAN REDUCE OZONE-PRECURSOR California’s standards for motor vehicle NOx emissions,
EMISSIONS set by CARB, are actually lagging the auto industry’s
capabilities. For example, one of GM’s largest passenger
THE IMPORTANCE OF NOX CONTROL sedans in production, the 1998 V-8 Mercury Grand
Marquis, achieves a NOx level of 0.004 g/mi. as certified
In a major study of urban and regional air pollution, the by the CARB (see Ref. 8). Yet CARB’s tightest NOx
National Research Council (NRC) concluded that “Of the standard is the 0.020 g/mi. SULEV (Super Ultra-low
six major air pollutants for which National Ambient Air Emissions Vehicle) which is five (5) times more lax.
Quality Standards (NAAQS) have been designated under Furthermore, the SULEV standard is not required until
the Clean Air Act, the most pervasive problem continues 2004 and then for only a small fraction of all vehicles.
to be [tropospheric (i.e., near-ground)] ozone, the most
prevalent photochemical oxidant and an important A biennial review of the ZEV Program is scheduled by
component of ‘smog’.” (NRC, Ref. 5) the CARB for September 2000.

Ozone forms when NOx combines with VOCs (volatile Vehicle NOx emissions can now be reduced to such low
organic compounds) in the presence of heat and levels that NOx emissions from both electric- and fueled-
sunlight. VOCs originate from a variety of sources both vehicles would be dominated by the up-stream fuel-cycle
anthropogenic (e.g., paints, vehicle exhaust emissions) emissions. Such emissions are directly proportional to
and biogenic (e.g., plants, trees). NOx, on the other hand, the vehicle’s efficiency in using electricity and fuel and
occurs almost entirely from high-temperature combustion are discussed in the next section.
sources, and these sources are largely anthropogenic
(e.g., vehicle engines, power plants). FUEL-CYCLE EMISSIONS DOMINATE TOTAL NOX
EMISSIONS
The NRC determined that NOx control would be key to
ozone control. “NOx control is necessary for effective When exhaust emissions are at SULEV levels (0.02
reduction of ozone in many areas of the United States. g/mi.) and below, total emissions will be dominated by
...in many urban cores and their environs, even if the emissions associated with the fuel cycle itself. Ref. 9
anthropogenic VOCs are totally eliminated, a high reports that, even for vehicles used in California, total
background concentration of reactive VOCs will NOx emissions (vehicle plus fuel-cycle) for gasoline and
diesel vehicles can be equal to, and even less than, total Energy Laboratory (NREL). Key results from this
fuel-cycle NOx emissions for electric vehicles. The data research are presented in Figure 4 and summarized in
reported in Ref. 9 were developed and reported in Ref. this section.
10 under a research contract for the National Renewable

Figure 4 - Total NOx emissions of a) 1998 Ford Expedition, b) 5 mi./kWh BEV, and c) four hypothetical HEVs that
would have the same 0.022 g/mi. NOx emissions in California as the BEV (all vehicles are operated in
the SoCAB in 2010, Scenario 4) (from Ref. 10)

29 mi/gal Diesel Vehicle, Tailpipe

29 mi/gal Diesel Vehicle, Tailpipe
NOx 0.01 g/mi.
NOx 0.01 g/mi. Total
TotalWorld
WorldNOx
NOxother
otherthan
thanUSUS
46.4 mi/gal Gasoline Vehicle,
46.4 mi/gal Gasoline Vehicle, Total US NOx except California
Tailpipe NOx 0.01 g/mi. Total US NOx except California
Tailpipe NOx 0.01 g/mi.
20.5 mi/gal Diesel Vehicle,
20.5 mi/gal Diesel Vehicle, Total
TotalCalifornia
CaliforniaNOx
NOxother
otherthan
thanSouth
SouthCoast
Coast
Tailpipe NOx 0.005 g/mi.
Tailpipe NOx 0.005 g/mi.
32.8 mi/gal Gasoline Vehicle,
32.8 mi/gal Gasoline Vehicle, Vehicle Tailpipe NOx (g/mi)
Vehicle Tailpipe NOx (g/mi)
Tailpipe NOx 0.005 g/mi.
Tailpipe NOx 0.005 g/mi.
Total NOx for 5 mi/kWh EV, Total
TotalSouth
SouthCoast
CoastFuel-Cycle
Fuel-CycleNOx
NOx
Total NOx for 5 mi/kWh EV,
Scenario 4
Scenario 4
13 mi/gal 1998 Ford Expedition
13 mi/gal 1998 Ford Expedition
(Gasoline) in Scenario 4
(Gasoline) in Scenario 4
NOx, g/mi. 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30
NOx, g/mi. 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30

1)1)Fuel-cycle
Fuel-cycledatadataarearesc aled proportionally
scaled proportionally(based
(basedupon
uponfuel
fueleconomy)
economy)for forSScenario
cenario44from
fromRef.
Ref.7,7,
Evaluation of Fuel-Cycle Emissions...,
Evaluation of Fuel-Cycle Emissions..., 9/19/96. 9/19/96.
2)2)Graph
Graphdepicts
depictsfour
fourhypothetical
hypotheticalvehicles
vehicleswith
withtotal
totalNOx
NOxemissions
emissionsininCalifornia
Californiathat
thatwould
wouldequal
equalthethetotal
total
NOx emissions in California of a 5 mi/kWh EV used in the
NOx emissions in California of a 5 mi/kWh EV used in the S oCAB in 2010. S oCAB in 2010.
3)3)InInthis
thisgraph
graphforforSScenario
cenario44inin2010,
2010,tailpipe
tailpipeNOx
NOxemissions
emissionsforforthe
thehypothetic al fueled
hypothetical fueledvehic
vehic les
lesare
are
assumed
assumed to be either 0.01 g/mi. or 0.005 g/mi. (For comparison purposes, in ARB tests, the full-size1998
to be either 0.01 g/mi. or 0.005 g/mi. (For comparison purposes, in ARB tests, the full-size 1998
Mercury Grand Marquis V8 w ith an advanced catalyst achieved
Mercury Grand Marquis V8 with an advanced catalyst achieved 0.004 g/mi.) 0.004 g/mi.)
4)4)Note
Notethat
thattotal
totalNOx
NOxemissions
emissionsininthe theUS
USfrom
fromthethe55mi/kWh
mi/kWhEV EVarearemore
morethan
thandouble
doublethe thetotal
totalNOx
NOx
emissions in the US from any of the hypothetical fueled
emissions in the US from any of the hypothetical fueled vehicles. vehicles.

In order to determine the magnitude of fuel-cycle primary anthropogenic contributor to ozone (as
emissions, the CARB contracted with Acurex Corporation discussed in the sections above).
for an analysis specifically aimed at emissions associated
with vehicles when they were operated in the Southern The Acurex study concludes that BEVs do offer a unique
California Air Basin (SoCAB). The purpose of the study advantage in the SoCAB, due to regulations capping NOx
is quoted directly from Ref. 7, “This study investigated emissions for SoCAB power plants. “Power plants in the
the emissions associated with the production and South Coast Air Basin are subject to the Regional Clean
distribution of conventional and alternative fuels. Air Initiatives Market (RECLAIM) regulation of the South
Emissions from the production and distribution of fuels Coast Air Quality Management District (SCAQMD),
are known as fuel-cycle emissions and these can be which provides a cap on power plant NOx emissions for
significant in comparison to tailpipe emissions....The each utility. For larger usages of EVs in 2010, power
California Air Resources Board’s (CARB’s) zero-emission generation will result in zero additional NOx in the South
vehicle (ZEV) classification is based on zero emissions Coast Air Basin due to RECLAIM limits.” (Ref. 7)
from electric vehicles; therefore, fuel-cycle emissions
associated with incremental power generation might be However, Ref. 9 concludes that for the rest of California
compared with incremental fuel-cycle emissions from and for the US as a whole, gasoline and diesel-powered
conventional fuels.” (Ref. 7) HEVs with even modest fuel efficiency may offer total
NOx emissions that are even less than the total NOx
Although the Acurex study (Ref. 7) looked at both NOx emissions that are associated with the use of BEVs in the
and VOCs, the research reported in this paper focuses SoCAB.
upon NOx emissions because of their importance as the
Figure 4 (from Ref. 9, 10) shows total NOx emissions for benefit of proper fuel tailoring tends to verify this claim.”
vehicles operated in the Southern California Air Basin (Ref. 8)
(SoCAB) when using diesel, gasoline or electricity. In
addition to the tailpipe NOx levels (which are zero for an The CARB was quite correct that BEVs result in low fuel-
EV), the graph shows the fuel-cycle NOx emissions cycle emissions in California (Ref. 8). In fact, in 2010
associated with each fuel-type and whether those almost three-quarters of all NOx emissions associated
emissions occur a) in the SoCAB, b) elsewhere in with the use of BEVs in the SoCAB (Southern California
California, c) elsewhere in the US or d) outside the US. Air Basin) will actually be produced in other states (see
(In Figure 4, Scenario 4 has been selected for Figure 4).
presentation; Scenario 4 represents additional efficiency
improvements over the Scenario 2 “standards applicable However, fueled vehicles could offer the same low NOx
in 2010.”) emissions in California. For example, the top two bars of
Figure 4 show a 29-mi/gal diesel vehicle and a 46 mi./gal
The bottom bar of Figure 4 depicts the emissions of the gasoline vehicle that have NOx tailpipe emissions of
1998 Ford Expedition, a 3,000-kg sport utility vehicle 0.010 g/mi.; both vehicles achieve the same California
(SUV). It’s 0.03 g/mi. NOx emissions at the tailpipe as NOx emissions (0.022 g/mi.) as the 5 mi./kWh EV.
determined by CARB testing are 0.01 g/mi. above the
ARB’s SULEV standard of 0.020 g/mi. (the tightest Furthermore, total US NOx emissions in the rest of the
standard for fueled vehicles). However, because of its US from these fueled vehicles would be roughly 15% that
relatively low fuel efficiency (13 mi./gal), its total fuel- of the BEVs (see Figure 4).
cycle NOx emissions from Figure 4 are more than 0.23
g/mi., of which 0.073 g/mi. are in California. About half Such fuel efficiency for HEVs in 2010 seems quite
(0.118 g/mi.) occur outside the US (e.g., drilling, tankers feasible given the actual performance of current high-
at sea). The rest (0.044 g/mi.) occur inside the US but mileage diesel- and gasoline-powered prototypes.
outside California.
The assumed NOx tailpipe emissions of 0.010 g/mi. also
The bar immediately above the Ford Expedition seem quite feasible in 2010; according to CARB (Ref. 8),
represents the fuel-cycle NOx emissions of a 5-mi/kWh even the 1998, 6-passenger Mercury Grand Marquis with
BEV. Such efficiency is deemed feasible for a variety of a large V-8 engine has already achieved 0.004 g/mi. NOx
small 2-4 passenger BEV designs in 2010. Note that tailpipe emissions.
while such a BEV would have total NOx fuel-cycle
emissions of almost 0.09 g/mi., only 0.022 g/mi. occur in The middle two bars in Figure 4 depict hypothetical
California. vehicles that have only 0.005 g/mi. NOx tailpipe
emissions (this is still 25% more NOx than the actual NOx
In the next section, HEVs are described that could have emissions of the 1998 Mercury Grand Marquis). If these
the same low 0.022 g/mi. California NOx emissions as the hypothetical vehicles achieved only 20.5 mi/gal (diesel)
5 mi./kWh BEV. or 30.8 mi/gal (gasoline), they, too, would produce the
same California NOx emissions as the 5 mi./kWh BEV.
HEVS USING FUELS OR ELECTRICITY CAN MATCH
THE CALIFORNIA FUEL-CYCLE NOX EMISSIONS These conclusions are especially important given the
FROM BEVS growing popularity of sport utility vehicles (SUV) with
their low fuel economy and correspondingly high fuel-
The CARB states that “ZEVs are the ‘Gold Standard’ cycle emissions. The CARB (1998) LEV II amendments
based upon their extremely low fuel-cycle emissions in to the 1990 LEV Program subject these vehicles in 2004
California and inherent emissions durability” (Ref. 8). and beyond to the same tailpipe standards as passenger
cars, and their is every indication that HEV versions of
Emissions durability of fueled vehicles was an important SUVs could achieve SULEV standards for tailpipe NOx
issue during the early 1990s, and the “inherent durability” emissions. Furthermore, HEV versions of SUVs also
of the BEV ECS was an appealing image. could achieve much better fuel economy and, therefore,
much lower fuel-cycle NOx emissions and CO2 emissions.
More recently, the CARB has been quite upbeat about
the durability of the ECS (emissions control system) of HEVS CAN BE A SOURCE OF ELECTRICITY FOR
fueled vehicles: “Discussions with catalyst suppliers THE GRID
indicate that with proper fuel tailoring adjustment, these
latest technology palladium-rhodium designs lose HEVs may not be just the next generation of mobility.
virtually none of their emissions conversions capability The concept of HEVs as a source of power to the grid
over more than 100,000 miles of aging. The small was first patented in 1980 (Ref. 11). More recently, the
increases in emissions of our two Expeditions without the participants in the Electric Technology Roadmap Initiative
(Ref. 12) have concluded that Distributed Generation HEVS CAN DISPLACE FUEL USE WITH GRID
st
(DG) will be a cornerstone of the 21 Century electricity- ELECTRICITY
production system; as part of a DG system, the grid not
only can deliver power to HEVs for recharging; the grid BEVs displace fuel use by using grid electricity. BEVs
can receive power from the same HEV. offer the greatest “gasoline equivalent fuel economy” as
defined by NREL (see Appendix A), but their limited
Already, stationary “co-generation” systems have been range, lengthy recharge time and high cost have limited
commercialized, and HEVs are often “stationary” for long their marketability to special niches.
periods of time. It is a relatively small step to envision
that valuable assets like HEVs may provide more of a HEVs that can be recharged overnight from the grid also
return-on-investment than just energy-efficient mobility. displace fuel use; when using grid electricity for the local
commutes and around-town trips, such HEVs can match
CONCLUSIONS the BEVs “gasoline equivalent fuel economy.” Such
HEVs also can travel unlimited long distances by using
HEVs can reduce fuel use (and therefore emissions of fuels.
CO2 and NOx) by improving fuel economy and by
displacing fuel use with utility electricity. In the longer term, as renewable energy technologies are
commercialized, HEVs that can recharge from the grid
HEVS CAN REDUCE FUEL USE, NOX AND CO2 BY may get their energy increasingly from renewable energy
IMPROVING FUEL ECONOMY sources, thereby further reducing both carbon emissions
and ozone precursor emissions (e.g., NOx).
“Of the six major air pollutants for which National
Ambient Air Quality Standards (NAAQS) have been HEVS CAN BE A SOURCE OF POWER TO THE GRID
designated under the Clean Air Act, the most pervasive
problem continues to be [tropospheric (i.e., near-ground)] HEVs that can be recharged from the grid may also be a
ozone… (NRC, Ref. 5). The NRC determined that NOx source of power to the grid. Analysis reported in Ref. 13
control was key to ozone control; “… even if indicates that such HEVs represent a sustainable
st
anthropogenic VOCs are totally eliminated, a high technology for the 21 century. These conclusions have
background concentration of reactive VOCs will been corroborated in Ref. 12, a major policy study from a
remain.... In the presence of anthropogenic NOx and broad coalition of participants led by the Electric Power
under favorable meteorological conditions, these Research Institute (EPRI).
background biogenic VOCs can contribute to
summertime ozone concentrations exceeding the MANY HEV DESIGNS ARE LIKELY TO FIND
NAAQS concentration of 120 ppb.” (Ref. 5) PROFITABLE MARKET NICHES

Vehicle NOx exhaust emissions can now be reduced to Toyota’s Prius HEV offers the range, performance and
such low levels that NOx emissions attributed to the fuel comfort of a CV, and it has been a major success in the
cycle dominate total vehicle NOx emissions. By improving Japanese market. Toyota is now offering a version in the
fuel economy, HEVs can reduce total NOx emissions US that is optimized for the American market. Sales of
(exhaust plus fuel cycle) to levels comparable to the fuel- the Prius have exceeded the combined sales of all BEVs
cycle NOx emissions attributable to electric vehicles. from all manufacturers. Honda has entered the HEV
market with the Insight. Although these entries are both
Improvements in fuel economy also reflect directly into a hybrids, they differ significantly in their design and
corresponding reduction in the production of CO2, a operation. Both are competitively priced.
principal greenhouse gas.
In a new industry like HEV technologies and in the
Fuelcell hybrid vehicles (FCV) offer much improved fuel absence of a definitive vision of the future, there can be a
economy when compared to conventional vehicles (CV). tendency for engineering designers to grasp prematurely
If technical and cost barriers can be overcome, FCVs for “the best” HEV design. The authors argue, on the
would be effective in reducing fuel use, thereby reducing other hand, that many different HEV designs can emerge
CO2 and fuel cycle NOx. that could meet regulatory requirements and find
profitable market niches. The future bodes well for those
In the near-term, HEVs using combustion engines offer in the ADVISOR community who can develop these
similar advantages. Such HEVs improve fuel economy by many diverse options for their marketing colleagues.
operating engines only in their most efficient domain, by
capturing and storing braking energy, and by optimizing
each vehicle subsystem and minimizing vehicle weight.
ACKNOWLEDGMENTS 10. Reuyl, J. S., “The impact of HEVs (hybrid-electric vehicles)
upon regional air quality, climate change and sustainable
economic growth,” January 22, 1999, from Reuyl, J. S.,
The authors express their thanks to NREL for the “Why HEVs could be considered ‘Super-ZEVs’ and be
opportunity to present this paper at the ADVISOR Users worthy of multiple credits,” research report for the National
Conference, August 24-25, 2000, Costa Mesa, Renewable Energy Laboratory (NREL) of the US
California. Special thanks are extended to Bob Kost at Department of Energy (DOE). October 28, 1998.
DOE and Terry Penney at NREL for sponsoring the 11. Reuyl, J. S., US Patent No. 4, 182, 960, “Integrated
Residential and Automotive Energy System,” January 8,
research reported in References 4 and 10. 1980.
12. “Electricity Technology Roadmap,” C1-112677-V1, July
CONTACTS 1999, Electric Power Research Institute.
13. Reuyl, John S., “Hybrid-Electric Vehicles, Energy and the
st
Environment - A Sustainable Scenario for the 21
NEVCOR, Inc. th
Century,” Invited Paper for the 17 Congress of the World
Dr. John Reuyl Dr. Robert Apter Energy Council, September 13-17, 1998.
PO Box 18683 28 Ave. 4-Marronniers
Stanford, CA 94309 1400 Yverdon
US Switzerland
Tel/Fax:1 650 701 0464 Tel/Fax: 41 24 426 7160
e-mail: [email protected] e-mail: [email protected]

REFERENCES

1. Reuyl, J. S., “Hybrid-electric Vehicles, Energy and the

Environment - A Sustainable Scenario for the 21st
Century,” invited paper for Division 4 - The Global Energy
Sector: Concepts for a Sustainable Future, 17th Congress
of the World Energy Council, Houston, September 13-17,
1998, pp. 143-152.
2. Apter, Robert and Edward Barnes, “Definition of engines
for hybrid applications,” Paper to be presented at SAE
Congress March 2001.
3. CARB (California Air Resources Board) (1990). “Proposed
Regulations for Low-Emission Vehicles and Clean Fuels,”
August 13, 1990, Staff Report.
4. Reuyl, John S. and Pierre J. Schuurmans, “Policy
Implications of Hybrid-Electric Vehicles,” under contract to
NREL, April 22, 1996, can be downloaded from
http://www.hev.doe.gov/papers/nevcor.pdf
5. NRC (National Research Council) 1991. Rethinking the
Ozone Problem in Urban and Regional Air Pollution,
Committee on Tropospheric Ozone Formation and
Measurement, National Academy Press, Washington, DC.
6. U.S. Department of Transport, Federal Highway
Administration, “1990 Nationwide Personal Transportation
Survey (NPTS) Public Use Tapes,” Volpe National
Transportation Systems Center, Cambridge, MA, 1991.
7. Acurex Environmental (1996), “Evaluation of Fuel-Cycle
Emissions on a Reactivity Basis,” September 19, 1996,
Final Report FR-96-114.
8. CARB (California Air Resources Board) (1998), “Proposed
Amendments to California Exhaust and Evaporative
Emissions Standards and Test Procedures for Passenger
Cars, Light-Duty Trucks and Medium-Duty vehicles ‘LEV
II’ and Proposed Amendments to California Motor Vehicle
Certification, Assembly-line and In-use Test Requirements
‘CAP 2000’”, September 18, 1998, Staff Report.
9. Reuyl, John S., “California Test Case: Success or are
Modifications Needed?”, Invited Keynote Paper for the
Transportation Track, Fifth International Conference on
Technologies and Combustion for a Clean Environment,
July 12-15, 1999, Lisbon Portugal.
DEFINITIONS, ACRONYMNS, ABBREVIATIONS SoCAB South Coast Air Basin
SULEV Super ULEV
Super-EV Super Electric Vehicle
ACV Advanced Conventional Vehicle Super-ZEV Super Zero-Emissions Vehicle
APU Auxiliary Power Unit SUV Sport Utility Vehicle
ARB (California) Air Resources Board ULEV Ultra-Low-Emissions Vehicle
BOEV Battery-only Electric Vehicle (BEV) VOC Volatile Organic Compounds
BEV Battery-Electric Vehicle (BOEV) VMT Vehicle Miles Traveled
CARB California Air Resources Board ZEV Zero-Emissions Vehicle
CNG Compressed Natural Gas
CYC_1015 Japanese 10-15 mode driving ADVISOR Simulation Parameters
cycle
ACC_CONV Accessories load, conventional
CV Conventional Vehicle vehicle
DG Distributed Generation ACC_HYBRID Accessories load, hybrid vehicle
DOE Department of Energy Ahr ampere-hour
ECS Emissions Control System CD Aerodynamic Coefficient
EPRI Electric Power Research Institute CS Control strategy
EV Electric Vehicle (BEV, BOEV) ESS_ Energy Storage System
FCV Fuel-cell Vehicle EX_ Exhaust system catalytic converter
FUDS Federal Urban Driving Schedule EX_SI Conventional converter for an SI
engine
HEV Hybrid-Electric Vehicle
EX_FUELCELL Null catalyst for fuel cell
HWFET Highway Fuel Economy Test
FA Frontal Area
kWh kilowatt-hour
FC_SI41 Fuel Converter, 1- l engine, 41 kW
LDV Light-duty Vehicle
FC_ANL50H2 Fuel Cell, 50 kW net
LEV Low-Emissions Vehicle
GC_PM32 Unique Mobility SR180
NAAQS National Ambient Air Quality Stds motor/controller
NEDC New European Driving Cycle MC_AC25 Solectria AC induction
NiMH Nickel-Metal Hydride motor/controller
NMOG Non-Methane Organic Gas PTC Powertrain control
NOx Nitrogen Oxides PTC_CONV 5-speed conventional-drivetrain
control
NPTS National Personal Transportation
Study PTC_FUELCELL Hybrid with thermostat CS
NRC National Research Council PTC_PAR 5-spd parallel electric-assist
NREL National Renewable Energy PTC_SER Series hybrid w/ pure thermostat
Laboratory CS
PEV Parallel Hybrid-electric Vehicle PTC_EV for electric vehicle
PM Particulate Matter TC_DUMMY Torque coupler, lossless belt drive
PNGV Partnership for a New Generation TX_1SPD Manual 1-spd, calling TX_VW for
of Vehicles losses
RECLAIM Regional Clear Air Initiatives TX_5SPD_SI Manual 5-spd, calling TX_VW for
Market losses
SCAQMD South Coast Air Quality VEH_SMCAR Road parameters for small car
Management District WH_SMCAR Wheel/axle for small car
SEV Series Hybrid-electric Vehicle Wt Weight, kg
SIP State Implementation Plan
APPENDICES

A - SIMULATION PARAMETERS FOR EACH VEHICLE TYPE <NAME.M FILES>

Simulation CV FCV SEV PEV BEV

Parameters
FC_ SI41_EMIS ANL5H2 SI41_EMIS SI41_EMIS -
EX_ SI FUELCELL SI SI -
TX_ 5SPD_SI 1SPD 1SPD 5SPD_SI 1SPD
WH_ SMCAR SMCAR SMCAR SMCAR SMCAR
VEH_ SMCAR SMCAR SMCAR SMCAR SMCAR
PTC_ CONV FUELCELL SER PAR EV
ACC_ CONV: 0 HYBRID: 0 HYBRID: 0 HYBRID: 0 HYBRID: 0
Wt, kg 830 1030 1030 930 1030
CD 0.37 0.37 0.37 0.37 0.37
2
FA, m 2 2 2 2 2
Wheel rad., m 0.27 0.27 0.27 0.27 0.27
ESS_ - Pb28 PB16_fund_Optima PB16_fund_Optima Pb28
MC_ - AC25 AC25 AC25 AC25
TC_ - - - DUMMY -
GC_ - - PM32 - -
ESS_PB16_FUND_OPTIMA.m: 12V, 16.5 Ahr, 6.68 kg module, 25-module pack
ESS_PB28.m: 12V, 28 Ahr 11.8 kg module, 25 module-pack

B - GASOLINE EQUIVALENT FUEL CONSUMPTION

From a communiqué from Tony Markel, NREL:

The gasoline equivalent fuel consumption for an electric vehicle is calculated as,

mpgge = (distance traveled, mi) / (energy used, J) * ( lower heating value of gasoline, J/g) * ( density of gasoline, g/l) /
(3.785 l/gal)

where “energy used” equals the integral of (the power out of the energy system + discharge losses) / (coulombic efficiency
over all discharge periods. The coulombic efficiency is accounted for here because you would encounter it in recharging
your system back to its original state.

For a hydrogen fuel cell,

mpgge = (hydrogen fuel economy, mpg) / (density of hydrogen, g/l) / ( lower heating value of hydrogen, J/g) * ( lower
heating value of gasoline, J/g) * (density of gasoline, g/l).

The density of hydrogen in our data file is that of compressed hydrogen stored at 24 Mpa (~3500 psi).
Implementing Optimization in ADVISOR Using VisualDOC
Presentation
Paper
John Garcelon and Valdimir Balabanov,
Vanderplaats Research & Development, Inc.

Detailed Vehicle Thermal Systems Modeling in ADVISOR

Through Integration with FLOWMASTER2
Presentation
Paper
Rory Lewis and Jason Burke, Flowmaster USA, Inc.
John
JohnGarcelon , Vladimir
Garcelon, VladimirBalabanov
Balabanov
Vanderplaats
VanderplaatsResearch
Researchand
and
Development,
Development,Inc.
Inc.
Leaders
LeadersininDesign
DesignOptimization
OptimizationTechnology
Technology
 Overview
■ Why optimize?
■ How does optimization work?
■ What is VisualDOC?
■ Previous Efforts - Coupling ADVISOR
and VisualDOC
■ New VisualDOC Features
■ VisualDOC API - Embedding VisualDOC
in ADVISOR
■ Summary
 Why Optimize?
■ Improve Performance

■ Gain Insight into Difficult Design Problems

■ Tool for designers

How Does Optimization Work?
 How Does Optimization Work?

■ Design Variables: {X}

– parameters that may change, inputs

■ System Responses
– objectives, obj(X): minimize / maximize
– constraints: gk(X) < 0
 Optimization Applications
■ Structures
■ Nonlinear Mechanics
■ Computational Fluid Dynamics
■ Simulation of Kinematics
■ Chemical Processing
■ Financial Planning
 What is VisualDOC?
■ General-Purpose Optimization System
– Add optimization to almost any analysis
■ Components
– Algorithm Library
– Response Program Interface
– Graphical User Interface
– Application Program Interface (API)
– Design Database
 VisualDOC Components
■ Algorithms
– Gradient-Based (MMFD, SLP, SQP)
– Response Surface Approximate Optimization
– Design of Experiments
■ Response Program Interface
– Couple VisualDOC to Analysis Programs
■ Design Database
– Leverage Design Studies
■ API
– Embedding Optimization
Previous Efforts: ADVISOR &
VisualDOC
MATLAB - Workspace 1
Text
ADVISOR File

MATLAB - Workspace 2
VisualDOC
ADVISOR
Text
File

■ Auto-size

■ Parallel and Series Control Strategy Design

 Control Strategy Design
■ Design Variables: Parallel Vehicles
– high and low SOC
– electric launch speed
– charge torque
– off torque and minimum torque fractions
■ Design Variables: Series Vehicles
– high and low SOC
– minimum off time
– charge power
– maximum and minimum power
– maximum power rise and fall rates
 Control Strategy Design

■ Responses
– Any combination of…
■ fuel economy
■ gradeability

■ emissions (HC, PM, CO, and Nox)

■ acceleration performance (0-60, 0-85, 40-60)

– As constraints and/or objectives

■ maximize, minimize, or meet a target
 Previous Results
■ Inverse of normalized fuel economy
 Issues!
■ Multiple MATLAB Workspaces
– each with > 20 megabyte memory hit
■ Difficult to alter existing problems and
setup new problems
■ Auto-size → Control Strategy
– Ignores interdependent design variables
■ Eliminates design flexibility
 New VisualDOC Features
■ Design Database
– Improves problem definition flexibility
■ Response Surface Approximations
– Extended range of application
■ fewer analysis
■ Design of Experiments
– Generation of irregular design spaces
– D-Optimal Design
– Mixed forward regression models
 New VisualDOC Features

■ Discrete Optimization
– Leverages design points in the database
■ Faster with fewer analyses
■ API
– Embed VisualDOC into ADVISOR
■ MATLAB calling structure
■ Direct access to the design database

– defining problems
– retrieving results
Embedding VisualDOC in
ADVISOR
Optimization
Problem
Definition

VisualDOC VisualDOC
ADVISOR
Optimization Database

Post-Processing
Design Database Concepts
Database
Interface
Inputs Responses

User
Etc
Control

A Task is a “snap-shot” of the

current problem definition (design Tasks Results
variables, responses, etc.) from the Task 1 Task 1
interface objects Task 2 Task 2
Results

Multiple tasks and their results New Task

may exist in a database

 Design Database
■ Object-Relational
■ Multiple Runs and Tasks
– DOE, Response Surface, Direct
■ Problem (task) definition
– Data, Algorithms
■ Results
– DOE, Response Surface, Direct
■ VisualDOC API
– Database Access Calls
Design Database Contents
■ Interface Objects
– Inputs, Responses, Auxiliary, Design
Control
■ Task Definition Objects
■ Results Objects
– Summary, Iteration, Design Point,
Approximations, DOE, Residual
 Embedding Tasks
■ Optimization Problem Definition

■ Calling Optimization Design Modules

■ Post-Processing Optimization Results

 Optimization Problem
Definition
■ How will users define design
optimization problems?
– ADVISOR GUI
– MATLAB command window
– VisualDOC GUI
■ Goal: Allow users to design almost any
ADVISOR variable using almost any
ADVISOR response
 Calling Design Modules
■ VisualDOC design modules accessible as
MATLAB function calls
– Inherently iterative

state = INITIALIZE;
while state != DONE
state = RSA( X, R );
Modify_ADVISOR_Variables( X );
R = Calculate_ADVISOR_Responses();
 Post-Processing Results
■ Database access functions
– Allow real-time monitoring of progress
– Complete response and design variable
histories
– Details about convergence, timing, errors
 Implementation Schedule
■ New ADVISOR release: August 2000
■ VisualDOC 2 release: September 2000
■ Prototype ADVISOR-VisualDOC API
– fall 2000
■ comparison tests with previous coupling
■ generation and testing of new optimization

problems using API

– winter 2000/2001
■ complete integration
 Summary
■ New VisualDOC API
– Easier, extendable optimization interface
for ADVISOR
– Native MATLAB function calls
– New algorithms
– Design Database
■ Apply design optimization throughout
ADVISOR
 Implementing Optimization in ADVISOR Using VisualDOC
John Garcelon
Vanderplaats Research and Development

ABSTRACT MATLAB, a static object library, and a text file interface.

Optimization techniques included gradient-based
Parallel hybrid electric vehicles (HEVs) offer the algorithms, response surface approximations, design of
potential, if properly designed, for significant fuel and experiments (DOE), discrete/integer optimization, and
emissions benefits as compared to conventional multi-objective optimization.
vehicles. This paper focuses on the application of
optimization tools (i.e. VisualDOC) via a new Application Gradient-based algorithms such as Modified Method of
Programming Interface (API) to the overall design of Feasible Directions (MMFD), Sequential Linear
hybrid electric vehicles through multi-variable and multi- Programming (SLP), and Sequential Quadratic
objective optimization using ADVISOR. Programming (SQP), all require gradients of the
objective and constraints to find a maximum or
INTRODUCTION minimum. Response surface approximate optimization
creates numerical approximations of the true responses
Any vehicle design is a complex endeavor that involves using a least square process. Response surface
multiple disciplines and multifaceted interacting systems approximations can significantly reduce the number of
(drive train, structures, aerodynamics, and auxiliary analyses required to find an optimum and do not require
systems). HEVs offer additional technical challenges gradients. Design of Experiments or statistical designs
because they incorporate relatively new and emerging provide an arrangement of design points (i.e., sets of
technologies. In most complex systems, it is difficult to design variable values) that are useful in exploring the
qualify how variables in the subsystems interact and design space. A DOE may be viewed as a sensitivity
influence responses. Parametric studies can offer a study or as a rational starting strategy for response
great deal of insight; however, as the number of design surface approximations.
variables increases, the interactions become less
tractable. Design optimization offers a powerful set of This paper describes the next version of VisualDOC and
tools that help engineers understand and improve design how it relates to ADVISOR. An earlier version of
options. VisualDOC was coupled with ADVISOR, and this paper
addresses how the new version will impact ADVISOR
Design optimization is the process of modifying specified users and developers.
parameters of a design in order to achieve specified
goals. Typically, the parameters to be modified are PREVIOUS EFFORTS
called design variables and the goals are called design
objectives. Most designs have some set of operating Early in 1999, Vanderplaats Research and Development
limitations; these are typically termed design constraints (VR&D) and the Systems Analysis Team at NREL,
[1]. coupled ADVISOR v2.2 [2] and VisualDOC v1.0 [3]. This
effort produced promising results and highlighted certain
Design optimization is an iterative process in that issues.
numerous analyses must be performed as the design
progresses from its nominal state to an optimum. The The most promising results from this effort were in the
primary advantage to using optimization techniques is areas of "auto-sizing" and control strategy optimization.
that the optimizer implicitly considers all interactions Both efforts were profitable with optimization identifying
between design variables. Therefore, as the design significant design improvements when applied to control
becomes more complex (i.e., we consider more design strategy optimization. The optimizer worked well with
variables), the optimization process automatically takes control strategy problems (serial designs with 8 design
into account these interactions and how they influence variables and parallel designs with 6 design variables).
performance. Furthermore, combinations of ADVISOR responses
could be considered as either objectives or constraints
VisualDOC [3] is a general purpose optimization system (i.e., fuel economy, gradeability, emissions, and
that provides the means to quickly couple numerical acceleration times). Other design problems were also
optimization to almost any analysis. Version 1.2 included investigated, most notably, gear ratio and shift strategy
a graphical user interface (GUI), an interface to optimization.
DESIGN STUDIES solution of two specific design problems: auto-size and
control strategy.
One of the big advantages of using ADVISOR and
VisualDOC was in the area of performing "design LIMITATIONS
studies." The design studies performed chiefly examined
the tradeoffs between fuel economy and emissions. In The design studies illustrated some major limitations in
these studies, numerous optimization runs were made to our coupling of ADVISOR and VisualDOC. From a
investigate if a sacrifice in fuel economy could development standpoint, the primary limitations were
significantly reduce emissions to EPA Tier II levels using ease of use and extensibility. To gain full advantage of
current test data as modeled in ADVISOR. reusing DOE runs in multiple optimizations and easily
altering the design study parameters, engineers needed
The following describes the general procedure taken in to use the VisualDOC GUI. Furthermore, it was very
these design studies: difficult to modify ADVISOR to consider even slightly
different design optimization problems.
1. Choose Vehicle Type.
2. Auto-size the Vehicle. This lack of flexibility limited the potential of what design
nd
3. Perform a 2 Order Koshal DOE of the control optimization may achieve in hybrid vehicle design.
strategy parameters. Considering powertrain control strategy optimization, the
4. Optimize the Control Strategy Using Response auto-size process determined a specific set of
Surfaces with the DOE Design Points. components based on the vehicle performance
constraints. Since the sizes of these components are
In performing these studies, we would modify the design dependent on the current control strategy parameters,
objectives and emissions constraints to gain insight into separating the design problem into two steps ignored
the problem. Since a DOE run only requires responses interdependencies.
(as opposed to objectives and constraints), a single DOE
run could be leveraged and used in multiple optimization Process performance also suffered. Figure 1 shows how
runs. Our DOE run required 28 ADVISOR analyses for a ADVISOR and VisualDOC worked when calling
parallel problem and the optimum was typically found in VisualDOC from ADVISOR. When ADVISOR called
less than 25 additional analyses. (NOTE: VisualDOC VisualDOC, a new copy of ADVISOR would start. The
uses an adaptive, nonlinear update of the response overhead required an additional 20 megabyte MATLAB
surface approximations, which is why the responses workspace. Furthermore, all information was transferred
surface approximate optimization requires analyses after via text files, which imposed significant constraints on
the DOE run.) Performing these design studies required the flexibility.
a formal coupling of ADVISOR and VisualDOC.
At this time, VR&D is developing the specifications for
COUPLING ADVISOR AND VISUALDOC the next version of VisualDOC. NREL engineers and
other clients needed more flexibility to embed
The coupling of ADVISOR and VisualDOC was done in VisualDOC into applications. Previous versions of
two ways. The distinction between them was in terms of VisualDOC and its predecessor DOC [4] were not
which GUI you used (i.e., ADVISOR or VisualDOC). designed for developers to embed its optimization
Because VisualDOC already supported a link with capabilities into applications. Instead, users would
MATLAB, NREL developed a script to run ADVISOR "wrap" VisualDOC optimization around an analysis
without its GUI (i.e., no_gui_vrd.m). Because program like ADVISOR. Thus, one of the important new
VisualDOC's GUI is a general-purpose optimization features of VisualDOC 2 would be an application
interface that most ADVISOR users would be unfamiliar programming interface (API).
with, the Systems analysis team provided integrated
access to VisualDOC via the ADVISOR GUI for the

MATLAB - Workspace 1
Text
ADVISOR File

MATLAB - Workspace 2
VisualDOC
ADVISOR
Text
File

Figure 1: Calling VisualDOC from ADVISOR

NEW VISUALDOC FEATURES ADVISOR have difficulty analyzing some of these design
points because they are at extremes. Furthermore, DOE
VisualDOC v2 is a significant enhancement to previous can easily create invalid design points (i.e., design points
versions. This section describes how these that are physically impossible). For example in
enhancements will influence ADVISOR. ADVISOR, you can provide a high and low state of
charge as control strategy parameters. The limits of
INPUTS AND RESPONSES these variables can overlap; therefore, when DOE
creates design points, it can easily create a point where
VisualDOC 2 only exchanges inputs and responses with the low state of charge is greater than the high state of
ADVISOR. Other optimization toolboxes require that one charge. By creating a constraint between these variables
objective and perhaps constraints be computed from the such that the low state can never be equal or greater
responses as the design variables change. VisualDOC than the high state, the DOE can enforce this constraint
simply exchanges inputs and internally converts when generating design points. The effect of this option
responses to objectives and constraints as required. is to reduce the total number of design points prior to
This abstraction allows for any combination of inputs and creating approximations and performing analyses.
responses in any problem without burdening ADVISOR
with scaling of design variables, constraints D-Optimal Designs
determination, and/or handling multiple objectives.
D-Optimal designs also reduce the number of design
DESIGN DATABASE points prior to running the analyses and generating the
approximations. The criterion that D-Optimal designs
Underlying version 2 is an object-relational database that use is to maximize the D-efficiency for a given number of
provides access to all VisualDOC results and problem design points. Recall that D-efficiency is calculated as
parameters. ADVISOR no longer needs text files to
transfer data with VisualDOC (see Figure 1). Instead, 100 T 1
D − efficiency = Z Z m
ADVISOR makes native MATLAB calls to the database n
to query, search, and sort design problem definitions and
results. The VisualDOC database is a binary, multi-user, where, Z is the model matrix (in least squares
platform independent database. The database provides terminology), n is the number of design points, and m is
flexibility and facilitates reuse of design data. This paper the number of terms in the approximation model [5]. The
discusses the database further in the section on the set of design points that maximizes the D-efficiency is
VisualDOC API. determined by an exchange algorithm [6].
RESPONSE SURFACE APPROXIMATIONS Mixed Forward Regression Models
Significant enhancements have been made in the Mixed forward regression models can have a big impact
response surface approximations. We have extended in reducing overall runtime when approximations need
the approximations using a design point filtering and only be generated once. Here, VisualDOC will create the
weighting approach. The least squares solver has also best approximations up to a second order polynomial for
been enhanced to reduce the influence of numerical a limited number of design points. This is important
noise and poor conditioning. When performing when one considers that a DOE and response surface
optimization runs in ADVISOR, users can typically approximations can become computationally expensive
expect faster convergence and fewer analysis. very quickly. In the case of a 5 design variable problem,
to generate a full second order polynomial approximation
DESIGN OF EXPERIMENTS requires 21 analyses; for a 10 design variable problem,
66 analyses are required; for a 15 design variable
The DOE module in VisualDOC has added features and problem, 136 analyses are required. Moreover, some
several new statistical designs. These are summarized statistical designs are unsuited for generating second
as follows: order approximations (e.g., two-level factorial designs).

• Enforced Design Variable Constraints The mixed forward regression model in VisualDOC 2 will
• D-Optimal Designs generate mixed order approximations to minimize
• Mixed Forward Regression Models estimating error for a limited number of analyses. Thus,
mixed second order approximations can be generated
Enforced Design Variable Constraints for a 10 variable problem with significantly less analyses
than the 66 analyses required for a full second order
Designers can easily create irregular design spaces by approximation.
defining design variable constraints and enforcing them
in DOE. All minimum variance designs implemented in
VisualDOC place design points at the limits of design
variables. As a result, most analysis programs including
DISCRETE/INTEGER OPTIMIZATION The foundation of the API like the VisualDOC system is
the database. This means that to use the API,
The branch and bound process [7] for discrete/integer developers must use the VisualDOC database. This is
optimization in VisualDOC has also been improved and not as constricting as it sounds since the database is
takes advantage of the design database to eliminate very efficient. For example, an empty design database
reanalysis of previous design points. This significantly consists of two files, the data and index files, and is less
reduces the computational cost of discrete problems than 15 kilobytes in size. Most average design runs (10
such as auto-size where the number of battery modules design variables and 30 responses) typically require a
must be an integer. total database size of approximately 50 kilobytes.
Furthermore, the database provides a convenient way to
The VisualDOC API is also a new feature in VisualDOC organize and store information that is already required.
2 and it is explained in the next section.
The database organizes design data and provides fast
VISUALDOC API access to VisualDOC parameters and results. The
database is multi-user and platform independent.
The VisualDOC API is designed to allow application Therefore, simultaneous ADVISOR runs may access the
developers to incorporate VisualDOC into their same database, or the VisualDOC GUI could even be
applications. The VisualDOC API is a departure from the used to post-process design runs from ADVISOR.
architecture of previous versions of VisualDOC where
VisualDOC was designed to "wrap" around an analysis The VisualDOC database has three sections, interface
code. Although this wrapping was easy to implement, objects, design task objects, and results objects.
there were two significant drawbacks. First, users Interface objects act like a white board for defining
needed to invest time understanding optimization and design problems. Design task objects are a snapshot of
the VisualDOC GUI. Second, the implementation was the interface objects that define a design optimization
problem dependent; users could not easily develop a problem, and the results objects are the outcome from
general optimization capability. running a design task. Figure 2 depicts the database
sections.
ADVISOR provides an intuitive, well-organized interface
to hybrid vehicle powertrain analysis. Removing that The process of taking a snapshot verifies that a task will
interface and replacing it with a general purpose run successfully. Multiple tasks in a single database
optimization interface made using VisualDOC with collects all runs and their results in a single place,
ADVISOR more difficult. Today, most applications making the design study process more convenient and
include a GUI and the VisualDOC API allows developers less prone to loosing information. The inverse operation
to leverage their GUI and analysis capabilities with the of loading design task data and restarting is also
optimization API from VisualDOC. Users can still easily available.
wrap VisualDOC 2 around an application; however, the
API provides even more flexibility to developers.

Database
Interface
Inputs Responses

User
Etc
Control

A Task is a “snap-shot” of the

current problem definition (design Tasks Results
variables, responses, etc.) from the Task 1 Task 1
interface objects Task 2 Task 2
Results

Multiple tasks and their results New Task

may exist in a database

Figure 2: Database Sections

The API consists of four main parts as follows: DESIGN MODULES

• Defining/Modifying Interface Objects. Design modules operate with design task objects to
• Functions That Invoke Design Modules. create results. Design modules encapsulate the
• Retrieving Result Objects. optimization algorithms. There are three primary design
• Utility Functions. modules. These are Direct Gradient-Base Optimization
(DGO), Response Surface Approximations (RSA), and
INTERFACE OBJECTS Design of Experiments (DOE).

ADVISOR primarily works with interface objects in the DGO supplies the MMFD, SLP, and SQP algorithms.
database. Interface objects consist of the following Discrete optimization is integrated into this design
object types: module and is automatically applied when required.

• Inputs RSA furnishes VR&D's latest response surface

• Responses technology. RSA runs can use several of the statistical
designs as a starting point or use a Taylor Series.
• Auxiliary
Discrete optimization is also integrated into this design
• Design Control
module. NOTE: Virtually any problem that RSA can run
can also be run using DGO and vice-versa. The switch
Input objects define what problem parameters may can be made by setting a simple flag in the design
change, i.e., design variables. There are over 18 control interface object.
different attributes for an input object that define
everything from the initial value and bounds on input to a
DOE gives users over 13 different statistical designs
name and description. The database provides default
along with multiple derivatives of these designs. DOE
values for all attributes. Virtually any variable in
can simply provide the design points or it can generate
ADVISOR may become an input object.
response approximations that can be used in
subsequent optimization runs in lieu of calling ADVISOR.
Responses are those values that ADVISOR calculates.
Responses may have limits and thus be considered
RESULT OBJECTS
design constraints. VisualDOC also allows you to specify
responses as design objects that you can minimize,
Each design module generates results while it runs. The
maximize, or direct towards a target value. There are
following result objects may be created by the different
over 24 attributes for each response object and the
design modules, depending on context:
database provides default values for all. These attributes
define limits on the response when used as a constraint
and how the response should be used as an objective if • Results Summary
the user desires so. • Iteration
• Design Point
Auxiliary objects are optional attributes that apply to • DOE
input and response objects. They allow for linked and • Approximation Models
synthetic inputs and responses and define discrete set • Residual
values for inputs.
The results summary object is continuously updated by
The design control object defines optimization algorithms all design modules and provides a convenient means for
and algorithm parameters. There are over 70 attributes ADVISOR to monitor the design run. There are over 35
that define what algorithms to employ and their different attributes that provide information about the
parameters. Everything from which statistical design to design module run including the best current objective
run for DOE to the finite difference step-size for value and worst constraint values.
calculating gradients is defined here. The database
provides default values for all attributes. Iteration objects provide a history of the design progress.
Each iteration object references a design point object.
A single API call, MakeTask, creates a design task from
the currently defined interface objects. MakeTask first Design point objects contain all design variable values
verifies that all interface objects are consistent and that and their corresponding response values. Design point
the design task will run. For example, every design task objects may be leveraged from one design task to
requires one or more inputs and responses. MakeTask another. VisualDOC 2 also leverages the corresponding
verifies that at least one input and one response have results. For example, ADVISOR may first make a DOE
been defined. The types of data verification checks run to generate design points and then use those points
made are in the context of the design task to be run. in multiple design studies by running RSA. Each RSA
Once the design task is created, ADVISOR may start run can use the design points (including responses) from
running a design module. the DOE run, thus reducing the computational
requirements and total time to perform design studies.
DOE objects are created for every DOE run. DOE
objects' attributes contain the statistical design MATLAB - Workspace
efficiencies and other attributes consistent among all
DOE design module runs. ADVISOR

The DOE design module also creates approximation

models objects whenever the DOE module generates VisualDOC
approximations. These attributes define the
approximating polynomials.

The DOE design module also generates residual objects Figure 4: VisualDOC in the ADVISOR workspace
for all responses when it generates their approximations.
The residual objects help measure the accuracy of the OPTIMIZATION PROBLEM DEFINITION
approximations. There are over 25 attributes for each
residual object. The question that needs to be answered here is how do
we want to present general optimization problems to
UTILITY FUNCTIONS ADVISOR users? The goal is to do this in an intuitive
manner that does not disrupt the continuity of the familiar
The utility functions of the API are used to open and ADVISOR GUI, but still easily allows for the definition of
create databases, check database integrity, and recover optimization problems in ADVISOR. There are a number
data from corrupted databases. Functions also exist to of options here that we are considering since it will have
query the version and build numbers for each module impact on both development and use. Since the
including the database. VisualDOC API allows us to define the problem either
via the ADVISOR GUI (i.e., m-functions), the MATLAB
Because each language including MATLAB has different command window, and even using the VisualDOC GUI,
function calling structures, the VisualDOC API provides we have a great deal of flexibility when prototyping this
language specific calls. The API supports native interface.
MATLAB calls; therefore, ADVISOR can directly call the
VisualDOC API from both m-files and the MATLAB VISUALDOC OPTIMIZATION
command window.
Once the VisualDOC database contains a design task,
COUPLING WITH ADVISOR ADVISOR initiates the optimization run. Because
VisualDOC's design modules are simply another
Coupling VisualDOC and ADVISOR using the API may MATLAB function call, the no_gui_vrd.m file no longer
be schematically described in Figure 3. Each of the three exists. The optimization iteration is directly coded within
arrows connected to ADVISOR define interface points.

Optimization
Problem
Definition

VisualDOC VisualDOC
ADVISOR
Optimization Database

Post-Processing
Figure 3: Coupling VisualDOC and ADVISOR

Before addressing the interface points, it is important to ADVISOR. For example, the RSA design module simply
realize that using the VisualDOC API will eliminate the alters design variable values and requests that
extra MATLAB workspace shown in Figure 1. Since ADVISOR supply response values for these new design
ADVISOR directly calls the VisualDOC design modules, variables. The following pseudo-code shows the basic
VisualDOC shares the ADVISOR workspace (Figure 4). premise.
 state = INITIALIZE; ACKNOWLEDGMENTS
while state != DONE
state = RSA( X, R ); The author would like to acknowledge the assistance of
Modify_ADVISOR_Variables( X ); the Systems analysis team in preparing this document
R = Calculate_ADVISOR_Responses(); and in performing the research described here. In
particular, the author would like to acknowledge the
efforts of Tony Markel.
Here, X and R are two MATLAB matrices. X represents
the current design variable values and R represents the CONTACT
corresponding responses that ADVISOR provides.
Modify_ADVISOR_Variables is a function that updates The author may be contacted by email at
the ADVISOR workspace to reflect the new design [email protected].
variable values, and Calculate_ADVISOR_Responses
computes the new responses according to the problem REFERENCES
definition.
1. Vanderplaats, G. N., Numerical Optimization Techniques
rd
for Engineering Design, 3 Edition, Vanderplaats
Because ADVISOR has many "dependent" variables Research and Development, Colorado Springs, CO, 1999.
(i.e., variables whose values dependent on other 2. ADVISOR 2.2 Reference, National Renewable Energy
variables), Modify_ADVISOR_Variables is not a trivial Laboratory, Golden, CO, October, 1999.
function nor is Calculate_ADVISOR_Responses. 3. VisualDOC 1.0 Reference, Vanderplaats Research and
However, we plan to use existing ADVISOR routines Development, Colorado Springs, CO, 1998.
with only minor modifications to generate the necessary 4. DOC 1.3 Reference, Vanderplaats Research and
Development, Colorado Springs, CO, 1995.
responses and to invoke the variable modifications for 5. Myers, R. H. and Montgomery, D. C., Response Surface
optimization. Methodology, Wiley Series in Probability and Statistics,
New York, NY, 1995.
POST-PROCESSING 6. Miller, A. J. and Nguyen, N-K., "A Fedorov Exchange
Algorithm for D-Optimal Design," Journal of the Royal
The Post-Processing connection is relatively easy. The Statistical Society, Volume, 43, No. 4, 1994.
7. Haftka, R. T. and Gürdel, Z., Elements of Structural
VisualDOC API provides functions for querying the Optimization, Kluwer Academic Publishers, Norwell, MA,
database as to the run status and providing intermediate 1993.
results for both inputs and responses. Real-time plotting
capability is available.

IMPLEMENTATION SCHEDULE

At the time of this writing, the next version of ADVISOR

is about to be released and VisualDOC 2 is in beta. The
API will be in the next major release of ADVISOR. By
early fall we expect to have a working prototype of the
system described here and be conducting comparison
tests with the old interface of ADVISOR and VisualDOC.

CONCLUSION

This paper describes current work being done to

incorporate the latest design optimization technology into
ADVISOR. Our goals in this work are to provide powerful
and flexible optimization capabilities to users of
ADVISOR such that they may optimize a wide variety of
hybrid vehicle design problems using the analysis
capabilities in ADVISOR.
 FLOWMASTER2

Rory Lewis and Jason Burke

Flowmaster USA, Inc.
What is ADVISOR?

• ADVISOR is a hybrid electric vehicle (HEV)

simulation model written in a widely used
software environment called MATLAB/Simulink.
It tests the impact of changes in vehicle
components, such as catalytic converters,
climate control systems, alternative fuels or
other modifications that might impact fuel
economy or emissions.
What is FLOWMASTER2?

• FLOWMASTER2 is a thermal-fluid network

simulation software package widely used by
automotive industry for the analysis of vehicle
thermal system performance.
Why ADVISOR & FM2?

• FLOWMASTER2 can easily model the details of

a vehicles thermal system. This would be
difficult to reproduce in ADVISOR.
• ADVISOR uses several factors, non thermal-
fluid subsystems, which would be difficult to
include in a FLOWMASTER2 model to
determine the boundary conditions in a vehicle
thermal system analysis.
Why ADVISOR & FM2?

• ADVISOR can provide FLOWMASTER2 with

better boundary conditions (e.g. engine heat
rejection, speed etc…)
• FLOWMASTER2 can provide ADVISOR with
better system temperatures and heat rejections
from the radiator and heater core.
• By integrating the two programs together, a
more powerful and comprehensive analysis can
be performed.
The ADVISOR model
The standard ADVISOR version 2.2.1 was used
for a conventional vehicle with an automatic
transmission:
• Conventional vehicle with FC_SI102_emis fuel converter,
1991 Dodge Caravan 3.0l;
• EX_SI exhaust aftertreatment;
• WH_SMCAR wheel/axle;
• VEH_LGCAR vehicle;
• ACC_CONV accessory.
• The data for vehicle mass, gear ratios, and wheel
diameter were changed to represent the 1991 Dodge
Caravan.
The ADVISOR model
The FLOWMASTER2 model

• The FLOWMASTER2 model used was

developed from previous large car system data.
The systems modeled include: engine coolant,
engine oil, transmission fluid, underhood airflow
and cabin airflow. These systems were
modeled from a thermal hydraulic basis.
The FLOWMASTER2 model
The ADVISOR - FM2 link

• ADVISOR controls the simulation

• FLOWMASTER2 runs in the background
• FLOWMASTER2 is called by a S-function in the
fuel converter sub model, fc_tmp, of the main
model bd_convat
• Microsoft COM/ActiveX capability in
FLOWMASTER2 and Matlab/Simulink allow
direct data exchange between the solvers
Data sent to FLOWMASTER2

Description ADVISOR Notes

Variable
Engine heat fc_th_pwr
rejection
Engine speed fc_spd_est
Vehicle speed mpha
Transmission loss gb_loss_fm not a standard variable
(as heat) in ADVISOR
Torque converter (htc_trq_in_a *
loss (as heat) htc_spd_in_a) -
(htc_trq_out_a *
htc_spd_out_a)
Ambient amb_tmp
temperature
Data sent to ADVISOR
Description ADVISOR Notes
Variable
Engine cylinder fc_tmp(1)
wall temperature
Coolant fc_clt_tmp & used for fuel economy
temperature at fc_tmp(2) and emissions
engine exit calculations
Underhood air fc_tmp(3) used in exhaust system
temperature heat transfer
Hood fc_tmp(4)
temperature
Oil sump fc_oil_tmp could be used for fuel
temperature economy and
emissions calculations
ATF sump gb_atf_tmp could be used in htc
temperature and gb loss prediction
Radiator heat fc_r_th_pwr
rejection
Cabin heater heat fc_h_th_pwr
rejection
The co-simulation

• The combined model was then run over the

Federal Urban Driving Schedule (FUDS) drive
cycle. This drive cycle lasts 1372 seconds and
is equivalent to the first two bags of FTP-75.
The Results
Exhaust Emissions (grams/mile)
Cold Hot Weighted % Error
Average
Advisor
CO 12.63 0.728 5.846 244%
NOx 1.785 0.738 1.188 116%
HC 1.981 0.328 1.039 352%
FLOWMASTER2 / ADVISOR Co-Simulation
CO 3.88 0.729 2.084 23%
NOx 1.63 0.738 1.122 104%
HC 0.807 0.329 0.535 133%
EPA Test
CO 1.7
NOx 0.55
HC 0.23
Cold Start Results with
Flowmaster2
Cold Start Results without
Flowmaster2
Conclusions

• Vehicle warm-up time significantly effects

emissions
• Accurately modeling the vehicle’s thermal
system is necessary to predict emissions
• An ADVISOR/FM2 co-simulation provides better
results then independent simulations
Questions
 Detailed Vehicle Thermal Systems Modeling in ADVISOR
Through Integration with FLOWMASTER2
Rory Lewis and Jason Burke
Flowmaster USA, Inc.

ABSTRACT With this combined model, the detailed effects on vehicle

performance, fuel economy and emissions by the
FLOWMASTER2 is a thermal-fluid network simulation thermal system can be examined. This combined model
software package widely used by automotive industry for can also be used to simulate complex drive cycle
the analysis of vehicle thermal system performance. boundary conditions on the vehicle thermal system.
The ability to integrate vehicle-specific thermal modeling These simulations can predict how the system will
with ADVISOR allows accurate prediction of the effects perform prior to road testing.
of thermal performance on the vehicle.
PROGRAMS
Every vehicle has different requirements for the control
of powerplant and passenger compartment FLOWMASTER2 is a One Dimensional internal fluid flow
temperatures. System configuration and individual analysis program. It consists of a single program, which
component performance has a significant effect on how is used to set-up, run and review a simulation. The
the vehicle thermal system performs. Through the solver uses linearized partial differential equations in a
integration of ADVISOR’s capability for complete vehicle matrix solution technique to solve for pressure, flow and
and drive cycle modeling with FLOWMASTER2’s temperature. FLOWMASTER2 can deal with steady and
capability of accurately modeling a vehicle’s entire transient flow, laminar and turbulent conditions and
thermal system, ADVISOR’s ability to predict heat flows incompressible and compressible flow. Hydraulic power
and temperatures throughout the vehicle will be transfer systems can be easily simulated. Heat transfer
improved. Thus, ADVISOR’s ability to predict emissions, can be modeled, including thermal inertia of the
fuel economy, performance etc. will be improved as well. surrounding structure. The general purpose solver can
solve linear, branching and looped networks. The
This integration will also benefit current FLOWMASTER2 Graphical User Interface is menu driven and has a
®
users through improved boundary conditions for the common Microsoft Windows look. Networks are drawn
vehicle thermal systems models. ADVISOR uses graphically on-screen and data entry, graph drawing,
several factors, non thermal-fluid subsystems, which results manipulation, results export and import, run
comparisons, custom reports and data security functions
1
to determine the boundary conditions in a vehicle are all supported.
thermal system analysis. Also, its built in ability to model
several standard drive cycles enhances the integration ADVISOR is a hybrid electric vehicle (HEV) simulation
between the two programs. These two factors combined model written in a widely used software environment
make ADVISOR an easy and user friendly way to called MATLAB/Simulink. This tool tests the impact of
incorporate this functionality into FLOWMASTER2. changes in vehicle components, such as catalytic
converters, climate control systems, alternative fuels or
Each program complements the other in terms of other modifications that might impact fuel economy or
capability, ease of use and functionality. By integrating emissions. The user can alter simulation results by
the two programs together, a more powerful and selecting vehicle component types, sizes and
2
comprehensive analysis can be performed. parameters. ADVISOR is developed and distributed by
the National Renewable Energy Lab (NREL).
INTRODUCTION
Simulink® is an interactive tool for modeling, simulating,
FLOWMASTER2 has been incorporated into ADVISOR and analyzing dynamic systems. Commonly used in
for the modeling of the vehicle thermal system through control system design, DSP design, communication
co-simulation. This replaces the standard model of the system design, and other simulation applications,
vehicle thermal system in ADVISOR for the conventional Simulink enables you to build graphical block diagrams,
vehicle model with an automatic transmission. The simulate dynamic systems, evaluate system
FLOWMASTER2 model was constructed using the performance, and refine your designs. Built on top of
FLOWMASTER2 graphical user interface and can be MATLAB®, Simulink offers immediate access to an
3
run and viewed directly through the GUI. The extensive range of analysis and design tools. MATLAB
appropriate Simulink sub-model in ADVISOR was and Simulink are developed and by The Mathworks.
modified to include an S-function that calls the
FLOWMASTER2 model.
MODEL The FLOWMASTER2 network models the flow of the
fluids through the system and the transfer of heat to the
The standard ADVISOR version 2.2.1 was used with the fluids from the powertrain and between the fluids.
following vehicle input: Conventional vehicle with Hydraulically independent fluids paths including loops
FC_SI102_emis fuel converter, 1991 Dodge Caravan and branches are used for the different fluids involved in
3.0l; EX_SI exhaust aftertreatment; TX_AUTO4 the vehicle thermal system. Fluid inertia and thermal
transmission; WH_SMCAR wheel/axle; VEH_LGCAR capacitance are modeled in each fluid stream. The heat
vehicle; PTC_CONVAT powertrain control; ACC_CONV flow through and thermal inertia of the systems solid
accessory. The data for vehicle mass, gear ratios, and components are also modeled. Energy, as heat, is
wheel diameter were changed to represent the 1991 added to the system from the engine to the coolant and
Dodge Caravan. The standard ADVISOR vehicle input engine oil, and from the torque converter and
screen with the data set as used can be seen in figure 1. transmission to the transmission fluid. System
components are modeled by either an individual
The FLOWMASTER2 model used was developed from FLOWMASTER2 component or, for greater detail, by a
previous large car system data. The systems modeled group of components. Each component is based on a
include: engine coolant, engine oil, transmission fluid, physical model with both geometric and performance
underhood airflow and cabin airflow. These systems parameters as inputs. The FLOWMASTER2 schematic
were modeled from a thermal hydraulic basis. The for the network used is shown in figure 2. Each of the
ADVISOR model provides system boundary conditions, different fluid systems is shown in a different color, with
operating conditions and thermal loading. the solid thermal model of the engine shown in red.

Figure 1: ADVISOR Vehicle Setup Screen

 Figure 2: Flowmaster2 Vehicle Thermal Systems Model
The following data is sent from ADVISOR to The combined model was then run over the Federal
FLOWMASTER2: Urban Driving Schedule (FUDS) drive cycle. This drive
cycle lasts 1372 seconds and is equivalent to the first
Description ADVISOR Notes two bags of FTP-75.
Variable
Engine heat fc_th_pwr
rejection
Engine speed fc_spd_est
Vehicle speed mpha
Transmission loss gb_loss_fm not a standard variable
(as heat) in ADVISOR
Torque converter (htc_trq_in_a *
loss (as heat) htc_spd_in_a) -
(htc_trq_out_a *
htc_spd_out_a)
Ambient amb_tmp
temperature

The following data is sent from FLOWMASTER2 to

ADVISOR: METHOD

Description ADVISOR Notes For the integration of the detailed FLOWMASTER2

Variable vehicle thermal systems model and the Simulink based
Engine cylinder fc_tmp(1) ADVISOR vehicle systems model, FLOWMASTER2 and
wall temperature Simulink were run together as a co-simulation. The
Coolant fc_clt_tmp & used for fuel economy ADVISOR GUI was used and Simulink controlled the
temperature at fc_tmp(2) and emissions simulation. FLOWMASTER2 was called from an S-
engine exit calculations function in the Simulink model.
Underhood air fc_tmp(3) used in exhaust system
temperature heat transfer During the initialization of ADVISOR’s conventional
Hood fc_tmp(4) vehicle model, FLOWMASTER2 is started and the
temperature appropriate model is loaded and initialized.
Oil sump fc_oil_tmp could be used for fuel FLOWMASTER2 then continues to run in the
temperature economy and background throughout the co-simulation. At each
emissions calculations ADVISOR time step, updated values of the co-simulation
ATF sump gb_atf_tmp could be used in htc variables are sent to FLOWMASTER2. The
temperature and gb loss prediction FLOWMASTER2 simulation then runs until it is at the
Radiator heat fc_r_th_pwr same time as ADVISOR. The variables are then read
rejection from FLOWMASTER2 into ADVISOR. ADVISOR runs
Cabin heater heat fc_h_th_pwr with a fixed one second timestep. The FLOWMASTER2
rejection model was run at a 0.25 second timestep, but can be run
at any timestep.
The Simulink model behind ADVISOR, bd_convat, was examined are for radiator heat rejection, emissions,
modified to call FLOWMASTER2. The bd_convat/fuel engine temperatures, and fuel economy. For the co-
converter <fc>/fuel use and EO emis/fc_tmp sub system simulation case, the radiator heat rejection and engine
contains the FLOWMASTER2 link S-Function. All temperatures are calculated in FLOWMASTER2 and
variables which are not standard inputs to this sub then sent to ADVISOR. Emissions and fuel economy
system are made available with Goto functions and read are calculated in ADVISOR, but are dependent on the
in with From functions. All of the outputs are standard results from FLOWMASTER2. The simulations were run
outputs for this sub system. for both hot and cold start cases. Figures 4 and 5 show
the results for the ADVISOR and co-simulation runs,
No changes have been made to the ADVISOR GUI, from a cold start.
therefore certain parameters must be set up in the S-
function prior to running the model, namely The results were also compared to the emissions
FLOWMASTER2 database and project directories, along certification information for the 1991 Dodge Caravan
with project and network names. from the EPA. The emissions results for the hot and
cold start cases were combined according to the
RESULTS weighted average used by the EPA in 1991, 43% of cold
start results and 57% of hot start results.
Results were generated with the ADVISOR-
FLOWMASTER2 co-simulation and with ADVISOR A comparison of the coolant temperature and heat
alone, using the same inputs. The results that were rejection at the radiator for the four cases are shown in
figures 6 and 7.

Figure 4: Co-Simulation Results (Cold Start)

 Figure 5: Advisor Simulation Results (Cold Start)
CONCLUSION

Exhaust Emissions (grams/mile) Differences in the vehicle thermal models can be seen to
Cold Hot Weighted % Error have an effect on the fuel economy and emissions of the
Average vehicle. The significant difference in the warm-up time
for the engine is the major reason for this. This is
Advisor
understood to be highly dependent on the thermal inertia
CO 12.63 0.728 5.846 244% of the thermal system, especially the volume of coolant
NOx 1.785 0.738 1.188 116% and the mass of the solids directly heated by the cooling
HC 1.981 0.328 1.039 352% system. Design and operation of the thermostat will also
FLOWMASTER2 / ADVISOR Co-Simulation effect the performance of the thermal system.
CO 3.88 0.729 2.084 23% For the hot start simulations, the emissions are seen to
NOx 1.63 0.738 1.122 104% be almost identical between the two cases. This is as
HC 0.807 0.329 0.535 133% expected, because as long as the thermal system can
EPA Test maintain a steady operating temperature, the emissions
of a known powertrain will not be effected by its design.
CO 1.7
Small fluctuations in the coolant temperature, due to the
NOx 0.55 lower system thermal inertia and detailed modeling of
HC 0.23 heat rejection, do not appear to have a noticeable impact
of the modeling of emissions. This detailed modeling is
valuable in that the simulation shows temperatures
above normal design maximums for cooling systems.
 ADVISOR Cold Start Co-Simulation Cold Start ADVISOR Hot Start Co-Simulation Hot Start

120.0

110.0

100.0

90.0
Coolant Temperature (C)

80.0

70.0

60.0

50.0

40.0

30.0

20.0
1 101 201 301 401 501 601 701 801 901 1001 1101 1201 1301
Time (s)

ADVISOR Cold Start Co-Simulation Cold Start ADVISOR Hot Start Co-Simulation Hot Start

40000

35000

30000
Radiator Heat Rejection (Watts)

25000

20000

15000

10000

5000

0
1 101 201 301 401 501 601 701 801 901 1001 1101 1201 1301
Time (s)

Figures 6 & 7: Comparison between ADVISOR and FLOWMASTER2 co-simulation results

However the emissions predicted are significantly higher
than those reported by the EPA for the vehicle. This
difference must be due to the input data used in the
simulation for the engines emissions.

The effect on vehicle fuel economy and emissions due to

the thermal system has been shown to be significant.
The use of simulation in the design of a vehicle that
includes accurate thermal system modeling will allow for
optimization of the thermal system for fuel economy and
emissions, not only to meet cooling and heating
requirements.

The ability to add a detailed FLOWMASTER2 vehicle

thermal systems model to ADVISOR increases the
flexibilty and accuracy of the complete vehicle and drive
cycle simulation.

ACKNOWLEDGMENTS

We would like to thank Brian McKay of Ricardo Inc. for

his work on the Simulink S-Function.

CONTACT

Rory Lewis, Flowmaster USA, 500 Davis St., Suite 504,

Evanston, IL 60201, [email protected]

1
Interfacing of 1D and 3D Fluid Dynamics Programs for the Simulation of a Diesel Engine Cooling System, Keith Austin,
Development Manager, Flowmaster International Ltd
2
www.ctts.nrel.gov/analysis/advisor.html
3
www.mathworks.com/products/simulink
 AAT
Vehicle Systems

ADVISOR Users Conference

Bob Kost
Vehicle Systems Team Leader
Office of Advanced Automotive Technologies
Office of Transportation Technologies
U. S. Department of Energy

Double Tree Hotel

Costa Mesa, California
August 24 - 25, 2000
 U.S. Highway Transportation Now Uses
More Oil Than Is Produced Domestically AAT
Vehicle Systems

18
16
14
Millions of Barrels per Day

12 Domestic Oil Production

10 Heavy Trucks

8 Gap

Passenger Vehicles
Light Trucks
6
4
2 Automobiles

0
1970 1980 1990 2000 2010 2020

Source: Transportation Energy Data Book: Edition 18, DOE/ORNL-6941, September 1998, and EIA Annual Energy
Outlook 1999, DOE/EIA-0383(99), December 1998
 Dramatic Increases in the use of
AAT
Trucks in the U.S.
Vehicle Systems

160,000

140,000

120,000

100,000
Automobiles
Thousands

80,000 Trucks
60,000

40,000

20,000

0
70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96
19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19
Years

Source: Stacy C. Davis, Transportation Energy Data Book, Edition 18, Sept. 1998, ORNL-6941
Note: FHWA data were used. Starting in 1993, some minivans and sport utility vehicles that were previously
included with automobiles were included with trucks.
 Future SUV? AAT
Vehicle Systems

Source: MH Designs, “The Future SUV”, The New SUV Standard, <http://www.poseur.4x4.org/futuresuv.html>,
(May 10, 1999)
 Office of Transportation Technology
Objective AAT
Vehicle Systems

❏ Improve fuel efficiency

Autos
Light trucks & SUVs

A More Energy
Independent
Increased
More Efficient Use of Nation
Vehicles
Alternative
Fuels
 OTT/OAAT Is Pursuing Broad Range of
Advanced Technologies AAT
Vehicle Systems
Energy Fuels
Conversion
•Gasoline/ Diesel
•CIDI •Natural Gas
•Fuel Cell •Hydrogen
•SIDI •Dimethyl Ether
•VCR •Ethanol
•Fischer-Tropsch Fuels
Energy
Management Advanced
•Batteries Materials
•Flywheels •Metals
•Ultracapacitors •Composites
•Ceramics
Power Powertrain
Configuration Other
Electronics Attributes
•Inverters •Parallel Hybrid
•Series Hybrid •Accessory Loads
•Motors
•Generators •Electric Vehicle
•Conventional
 Government/Industry Working
Together Under PNGV AAT
Vehicle Systems

Suppliers
Universities
Small Business
Daimler Federal Labs
Ford GM
Chrysler DOC DOE NSF DOT
Capabilities Technologies

Prioritized Needs Government Technologies

Industry Government
Resources Partnership Resources Agencies
USCAR Technologies (DOC Lead)
(PNGV)

Goal 1:
Adv. Manufacturing
DOD EPA NASA
Goal 2:
Near-Term Vehicle
Improvements
Goal 3:
Triple Fuel Efficiency
(up to 80 mpg)
 OAAT R&D Plan: Strategy
“Systems Driven - Barrier Focused” AAT
Vehicle Systems

❏ Derive all technical targets from a

Vehicle Systems Common Vehicle System Perspective
Driven* ❏ Culminate efforts with technology
validation at the Vehicle System Level

❏ Concentrate available funding on the most

Barrier-Focused critical technical barriers to successful
technology development (Most “Bang for
The Buck”)
*R&D Constraints
– Emissions Control Regulations (projected to be in place when technology is
available for the marketplace)
– Safety Standards
– Attributes of comparable, competitive vehicles (including cost)
 Office of Advanced
AAT
Automotive Technologies
Vehicle Systems
Vehicle Systems

Bob Kirk Director

Ed Wall, Laura Porterfield,

PNGV Coordinator Resources Coordinator

Vehicle Energy Energy

Systems Conversion Management
Bob Kost Steve Chalk Ray Sutula
Technology Requirements Fuel Cells EV Batteries
Definition & Validation Piston Engines High-Power Energy
Technology Development Fuels Storage
Lightweight Materials Prop. Sys. Materials Power Electronics &
Electric Machines

Technology Integration (Cross cut)

 Vehicle Systems Technology
AAT
Objectives
Vehicle Systems
Vehicle Systems

Phase 1: By 1998, develop and validate production feasible propulsion

subsystem technologies that will enable the achievement of 50 mpg in
test-bed six passenger sedans that meet EPA Tier 2 emissions and
retain all attributes and features of competitive automobiles

Phase 2: By 2004, develop and validate propulsion subsystem

technologies and validate OAAT developed technologies that will
enable the achievement of 80 mpg in six passenger sedans etc.

Phase 3: By 2011, develop and validate production feasible vehicle

system technologies that will enable achievement of 100 mpg in six-
passenger sedans emphasizing non-petroleum fuels and zero emissions
 Phase 1: Serious Investment by
AAT
Government & Industry
Vehicle Systems

Ford P2000

$116M* $151M*
Chrysler ESX

* 50/50 cost
share $52.6M*
 Methodology for Managing Vehicle Systems
Consists of Three Integrated Activities AAT
Vehicle Systems
Vehicle Systems

Systems Modeling &

Analysis
Digital Vehicle Hybrid Test Facility
• Guide/Prioritize Future R&D
•Sets Requirements & Targets
• Allows rapid layout to view •Predicts Performance •Model Validation
problems and opportunities (F.E. Emissions, Transient, etc.)
•Validation of Component &
•Helps OEM suppliers to •Control Strategy Development Subsystem Technologies
become better prepared for •Component & Subsystem •Benchmark technologies &
system level designs Model Development vehicles worldwide
•Fewer design iterations & • Test Procedure Development •Component, Engine &
faster convergence on solution Vehicle Characterizations
• Vehicle Models
•Ties together many different (SUV, Lt Truck, HD, Auto) •Hardware-in-the-loop
component projects in systems
context for minimum energy • Optimization Techniques •Controls Strategy Development
and emissions • Evaluate new Concepts for improved efficiency & lower
emissions
•Gives 1st Order Vehicle • Parametric Studies
Designs & “packageable” •Integration Tech. Development
• Cost/Benefit Analysis
products •Test Procedure Development
 Advisor & Digital Functional Vehicle Linkage
AAT
ADVISOR Digital Functional Vehicle
Vehicle Systems

■ Integrates Data from Many Different

Vehicle Performance
Analysis/Design Software
Optimization
■ Allows Thermal Effects Studies
Cycle Analysis
■ Passenger Comfort with min energy
Fuel Economy
■ Makes Energy Optimization Integral to
Emissions
Vehicle Development
■ Ties all phases of vehicle development
to a common model/process
 Migrating Toward Digital Functional Vehicle:
Ultimate Goal AAT
Vehicle Systems

❏ Lower cost than building hardware -- more effective use

of taxpayer money

❏ Allows trade-offs to be done earlier in design process

❏ Useful for setting future R&D directions

❏ Allows system level optimization with a focus on energy

and emissions
Digital Functional Vehicle Process Highlights
Energy/Emissions Impacts AAT
Vehicle Systems
 Results from DOE’s
Transportation R&D Are Displayed AAT
at International Auto Shows
Vehicle Systems

2000 Washington DC 2000 Detroit Auto Show

Dodge ESX3
- Body system weighs 46% less* Ford Prodigy Concept Vehicle
- Efficient diesel engine, motor and battery achieve 72 mpg* - Lightweight materials reduce vehicle weight 30% *
- Cost penalty halved to $7500 - Integrated starter/alternator *
- 33% reduction in aerodynamic drag
- Advanced diesel engine with 35% efficiency
2000 Detroit Auto Show improvement *
- High power battery *

GM Precept Concept Vehicle

- Vehicle mass reduced 45% *
- Eliminates need for power steering
- Lowest drag coefficient ever recorded for a 5-p sedan
- Fuel cell version achieved 108 mpg *
*DOE supported technologies
 Future Challenges AAT
Vehicle Systems
Vehicle Systems

•Continue to Develop Tools and Processes for Systems

Integration and Optimization

•Provide Web Based Tools for Automotive Suppliers

• Develop Climate Control and Thermal Models

• Develop Engine Emission and After-treatment Control

Models

• Develop System Cost Benefit Analysis Tools

 Simulation and Validation of Hybrid Electric Vehicles
using PSAT and PSAT-PRO
Presentation
Paper
Aymeric Rousseau and Maxime Pasquier, Argonne National Laboratory

The Reverse Engineering of a Diesel Engine:

A Unified Systems Approach Using ADVISOR
Presentation
Paper
George Delagrammatikas and Dennis Assanis, University of Michigan
 Using Forward Modeling of
Hybrid Electric Vehicles with
PSAT and PSAT-PRO

Maxime Pasquier ([email protected]),

Aymeric Rousseau ([email protected]),

ADVISOR Conference, Costa Mesa, CA, August 24-25

Argonne National Laboratory

1 Transportation Technology R&D Center
 ADVISOR
Ø Well-developed, powerful GUI, quick run-time
Ø Backward model

PSAT
Ø Control-oriented, transients, many configurations
Ø Forward model

Argonne National Laboratory

2 Transportation Technology R&D Center
 Differences Between Backward-Looking and
Forward-Looking Models
Backward looking model: Operation comes from cycle, components are not controlled
The desired speed is used to estimate operation from one component to another one

Forward looking model: Modeling with commands

Commands from a Powertrain Controller obtain the desired vehicle speed

Argonne National Laboratory

3 Transportation Technology R&D Center
 Forward-Looking Model Capabilities

Ø Model and test:

– Detailed powertrain component models,
– Control strategies can be tested using real
components and implemented in a vehicle micro-
controller.
Ø Model the powertrain’s transient phases ( engine
starting, shifting, clutch engagement /
disengagement…)

Because the modeled components behave as in reality, we have

1) A higher precision of the simulation results
2) The possibility to test and validate the models on the test stand
using prototyping
Argonne National Laboratory
4 Transportation Technology R&D Center
 ANL’s Systems Analysis Program

MODELING (PSAT): PROTOTYPING (PSAT-PRO):

ØChoose the appropriate powertrain ØIntegrate the strategy in a VCU
ØChoose component size ØSimulate in RT the test procedures
ØDevelop the best strategy ØControl the actual vehicle

TESTING (APTF):
ØPerform the simulated tests
ØTest and map components
ØTest hybrid vehicles

Argonne National Laboratory

5 Transportation Technology R&D Center
 What is PNGV System Analysis Toolkit (PSAT)?

A powerful modeling tool that allows the user to realistically

simulate:
ØFuel consumption and exhaust emissions (eg.
Federal Test Procedure, highway, all other cycles)
ØPerformance (eg. 0-30mph, 0-60 mph, 40-60 mph,
distance in 5 sec., maximum launch grade, maximum
continuous speed, 55mph at 6% grade)

Argonne National Laboratory

6 Transportation Technology R&D Center
 PSAT History

Ø First developed by Southwest Research

Institute (SwRI) from 1995 to early 1999
Ø Program transferred to ANL in September
1999
Ø First funded by USCAR and now by DOE
Ø Developed under the direction and with the
contributions of Ford, GM and
DaimlerChrysler

Argonne National Laboratory

7 Transportation Technology R&D Center
 Common GUI Development

Ø DOE’s goal for PSAT and ADVISOR to share

common GUI
Ø Since potential users are familiar with ADVISOR
GUI, common GUI will operate like current
ADVISOR GUI
Ø PSAT-specific functionality has been added to
current ADVISOR GUI

Argonne National Laboratory

8 Transportation Technology R&D Center
 Common GUI Integration

Argonne National Laboratory

9 Transportation Technology R&D Center
 PSAT Currently Includes Many Configurations

Ø Large number of drivetrain configurations:

– 3 conventional vehicles

– 48 parallel hybrids

– 24 series hybrids

– 24 fuel cell hybrids

– 1 power split hybrid (Prius-like)

A wide variety of configurations is needed because

we still don’t know which one is the best for specific
applications !!

Argonne National Laboratory

10 Transportation Technology R&D Center
 PSAT V3.0 Has Modular Component Models

ANL - PSAT V3.0

nomenclature
ØAll models follow Bond Graph principle
Command from Info to Controller
Controller ØAll models have consistent input/output
nomenclature
ØComponent models are plug-and-play
Effort MODEL Effort ØBlock diagram code representation for
Flow Flow simple visualization

Effort = Torque or Voltage

Flow = Speed or Current

Argonne National Laboratory

11 Transportation Technology R&D Center
 Parallel HEV SIMULINK Diagram in PSAT V3.0

Argonne National Laboratory

12 Transportation Technology R&D Center
 PSAT V3.0 Flows Intuitively

Accelerator/Brake pedal

Motor command

Engine command Shift command Brake command

Clutch command

Engine Mech. acc Clutch Transmission Final drive Wheel Vehicle

We can have 4 different positions

for the motor

Battery Elec. acc Motor Reduction

Argonne National Laboratory

13 Transportation Technology R&D Center
 PSAT Main Capabilities Summary

Ø Due to forward-looking philosophy:

– Model reality with real commands,
– Test advanced component models and control strategies,
– Take into account transient phases.
Ø Due to ANL’s work:
– Each drivetrain is built according to the user choices
– Large number of drivetrain configurations,
– Easy integration of new models, data or control strategies,
– Better organization allowing us to facilitate the link with the
Prototyping phase.

Argonne National Laboratory

14 Transportation Technology R&D Center
 ANL’s Systems Analysis Program

MODELING (PSAT): PROTOTYPING (PSAT-PRO):

ØChoose the appropriate powertrain ØIntegrate the strategy in a VCU
ØChoose component size ØSimulate in RT the test procedures
ØDevelop the best strategy ØControl the actual vehicle

TESTING (APTF):
ØPerform the simulated tests
ØTest and map components
ØTest hybrid vehicles

Argonne National Laboratory

15 Transportation Technology R&D Center
 Prototyping Definition

Simulated Actual
Vehicle DOWNLOAD Vehicle
Controller Controller
measures measures
commands
commands

Simulated MODELING Actual

Vehicle Vehicle

To integrate a controller in an actual vehicle,

extensions to PSAT are needed for prototyping:
Argonne created PSAT-PRO
Argonne National Laboratory
16 Transportation Technology R&D Center
 PSAT-PRO Features

Purpose: Users can go from PSAT modeling to prototyping

using PSAT-PRO for any kind of vehicle or configuration

Generic and reusable

Linked to PSAT

Three steps used to control a powertrain

Provides model validation

Argonne National Laboratory

17 Transportation Technology R&D Center
 PSAT-PRO Is Generic and Reusable

Test Procedure: We send the commands to follow the

desired test procedure.

Control Command System: We command each

component to follow the
test procedure in
accordance with the control
strategy and with the
components constraints.

Physical System Model: The test stand or vehicle

model should react exactly
like the real system.

Argonne National Laboratory

18 Transportation Technology R&D Center
 PSAT-PRO Is linked to PSAT

Driver

Check State Parameters Controller developed in PSAT Saturation E-stop

Dynamometer

Physical System Model

Argonne National Laboratory

19 Transportation Technology R&D Center
 PSAT-PRO Is linked to PSAT

Driver Test Procedure

Control Command System

Dynamometer

Physical System Model

Argonne National Laboratory

20 Transportation Technology R&D Center
 PSAT-PRO 3 Steps Are Used To Control a Powertrain

1-Simulation :
We simulate the test procedure with the
Test Procedure
Physical System Model composed by
the components library in order to check
the model.

2-Simulation real-time (HIL):

Control Command We simulate the test procedure in real-
System time with the Physical System in order to
check the control command system.

3-Control real-time (Rapid Prototyping):

We command the components to follow
the test procedure and we control the
Dynamometer to represent the vehicle
behavior.
Physical System Physical System Model

Argonne National Laboratory

21 Transportation Technology R&D Center
 PSAT-PRO Test Methodology Provides A Validated Toolkit

Test
Procedure
Real World
Simulation

Control
Command dSpace
System

Analysis of Measured
Simulation Prototype
Physical Differences Results
Results Drivetrain
System
Model

Model Modification

Argonne National Laboratory

22 Transportation Technology R&D Center
 PSAT-PRO ANL Multi-step Development / Validation Plan:

Phase 1 : Motor Phase 2 : Engine (with clutch and starter)

clutch

Motor Eng
Dyno Dyno
45 kW 1.7 L

Phase 3 : Conventional Vehicle with CVT Phase 4 : Post-Trans Parallel HEV with CVT

Dyno Dyno
Eng Nissan 4:1 4:1
Eng Motor Nissan
1.7 L CVT 1.7 L CVT
45kW

Modeled vehicle Modeled vehicle

Argonne National Laboratory

23 Transportation Technology R&D Center
 PSAT-PRO Main Capabilities Summary

Ø High link between PSAT and PSAT-PRO

Ø As PSAT, PSAT-PRO is generic and reusable
Ø Any drivetrain configurations can be tested
Ø Easy integration of control strategies developed in PSAT
Ø PSAT can be validated as we use the same component models
Ø Possibility to simulate in real time to analyze the differences between
simulation results and measures
Ø Post-processing tools facilitate the comparison simulation/test.

Argonne National Laboratory

24 Transportation Technology R&D Center
 PSAT and ADVISOR Conclusion

Ø ANL and NREL are working together to provide tools for

industry and academia for HEV systems analysis
Ø PSAT now has, ADVISOR will soon utilize common
variable naming convention
Ø Common GUI for both models will soon be available
Ø Test data from ANL is used for both models

Argonne National Laboratory

25 Transportation Technology R&D Center
 Users of PSAT Conclusion

ANL would like to work with you if you:

Ø are interested in developing controls or wish to use
modeling for prototyping,
Ø have detailed, forward-facing component models that you
would like to incorporate in PSAT,
Ø want to optimize a control strategy and test it in a vehicle

Argonne National Laboratory

26 Transportation Technology R&D Center
 Simulation and Validation of Hybrid Electric Vehicles using
PSAT and PSAT-PRO
Aymeric Rousseau and Maxime Pasquier
Argonne National Laboratory, 9700 South Cass Avenue, Argonne, IL 60439-4815

ABSTRACT The Partnership for a New Generation of Vehicles

(PNGV) is an historic public/private partnership between
ADVISOR is a user-friendly, publicly available model the U.S. federal government. The PNGV, which is led
useful in making fuel economy, performance, and by the Technology Administration at the Department
emissions prediction in vehicles. ANL and NREL are of Commerce (including 7 agencies and 19 federal
working together to make a suite of HEV technology laboratories) and DaimlerChrysler, Ford, and General
simulation tools the developer can use to help design Motors, aims to strengthen America’s competitiveness
HEV technology in a systems context. ADVISOR is an by developing technologies for a new generation of
easy to use vehicle simulation tool that has a wide user vehicles. The PNGV’s long-term goal, dubbed the
base. ANL is continuing the development of the PNGV “Supercar” goal, is to develop an environmentally
Systems Analysis Toolkit (PSAT) software model. A friendly car with up to triple the fuel efficiency of
version consisting of non-proprietary component models today’s midsize cars without sacrificing affordability,
is now available to component and systems researchers performance, or safety. The other two PNGV goals are
and developers. Both PSAT and ADVISOR have been to significantly improve national competitiveness in
modified to better exchange component data by sharing automotive manufacturing and to apply commercially
many of their variable names. The NREL-developed viable innovations to conventional vehicles.
Graphical User Interface (GUI) familiar to ADVISOR
users can now be used for PSAT after being modified. The most promising design to increase fuel efficiency
The GUI makes it easy to do component sizing, test and decrease emissions seems to be Hybrid Electric
execution, optimization, and visualization. PSAT, a Vehicles (HEVs). HEVs are vehicles that have both
forward-looking model, is well suited for development of electric and fuel-consuming power sources [Combes and
control strategies and writing accurate dynamic Cottard, 1992].
component models because code can directly be
imported and tested on a bench or in a vehicle. This In a world of growing competitiveness, the role of
paper will describe the ways vehicle the simulation simulation in vehicle development is constantly
models can be used to develop control systems by increasing. Because of the number of possible hybrid
creating a seamless bridge between simulation and architectures, the development of this new generation of
testing. This method is called the “Mechatronic vehicles will require accurate, flexible simulation tools.
Approach,” a powerful approach to rapid design and Such a simulation program is necessary to quickly
validation. ANL uses a Hardware-In-the-Loop (HIL) narrow the technology focus of the PNGV to those
control unit based on the dSpace real-time computer configurations and components that are best suited for
control system to link individual components or full achieving these goals. Therefore, the simulation should
powertrain systems to the PSAT code. These HIL tests be flexible enough to encompass the wide variety of
allow validation of control laws and physical system components and drivetrain configurations. Finally, it
response. The results of the HIL tests help refine the must be able to assist vehicle designers in developing
models, which in turn allow for more accurate simulation specific strategies and implement them on prototypes.
studies. The HIL system can also be implemented in
controlling an entire powertrain in a mule test vehicle for In order to respond to the needs of industry, Argonne
chassis dynamometer validation. The possible National Laboratory (ANL) undertook a collaborative
configurations and test objectives are outlined with effort to further develop the PNGV System Analysis
examples and figures from Argonne’s powertrain test Toolkit (PSAT) under the direction and contributions of
laboratory. Ford, GM, and Daimler-Chrysler. The model architecture
is “forward-looking,” meaning that component inter-
actions are “real world.” This method is computationally
INTRODUCTION more intensive than “backward-looking” architecture;
however, the result is a tool that will allow the advanced
Growing environmental and economic concerns have led powertrain designer(s) to develop realistic control
the U.S. government to impose new emissions control strategies and assess component behaviors in a system
regulations and consider new requirements for fuel environment by using models closer to reality. These
efficiency by car manufacturers. models were developed by using Matlab v5.3 and
Simulink v3. A nonproprietary version of this software is PSAT PRESENTATION
expected to be released in 2000.
To run a simulation, the user will have to first define
To respond to these attempts (modeling and validation some choices, such as the type of drivetrain or
of HEVs using Hardware in the Loop), the mechatronic component. Then, PSAT gives the user the choice to
approach has been used [DeCharentenay et al., create his/her own components (engine, motor, battery,
1996]. Mechatronics represents a new generation of transmission) by scaling existing data. Finally, for the
products that bring together elements of mechanical parallel and series configurations, the user can chose if
engineering and electrical and electronic technologies he/she wants to have a fixed ratio in between the motor
with information technology and software engineering. and the principal power output shaft.

This approach has been applied to the HEVs by using As PSAT is able to run both performance and
the bond graph methodology. Bond Graphs [Karnopp consumption/emissions tests, the user should also
et al., 1990] are graphical descriptions of dynamics choose what type of test should be done. In the case of
models based on power and informations flows. This an energy consumption test for hybrid configuration, a
technique offers the unique particularity to explicitly State-Of-Charge (SOC) equalization algorithm is also
describe not only the energetic exchanges between the available so that the consumption results of different
base elements of the physical structure of the system, configurations or strategies can be compared with same
but also the structure of its calculation. Moreover, the SOC.
Bond Graph, as it uses a pictorial form of the physical
structure and the calculated structure, appears as a link According to the user’s choices, the software will build
between the perception of the physical studied system, the appropriate model by using the right models,
its models, their exploitation on a microcontroller, and powertrain controller, and initialization files. The model
the interpretation of the results. will then be run automatically, and the results will be
provided in the Matlab prompt command.
In order to validate the models and their commands, we
use the Hardware-In-the-Loop (HIL) method with a CONVENTIONAL CONFIGURATION
Dspace 1103 board. This approach allows us to test and
validate the components and the whole drivetrain on the The main difference among conventional configurations
test bench. is the choice of transmission: manual, automatic, or
CVT. These drivetrains are used to validate PSAT’s
The main objective of this paper is to present the results with existing vehicles and serve as reference.
possibilities and characteristics of PSAT. We begin with
a description of the different drivetrain configurations that HYBRID CONFIGURATIONS
can be modeled. We then describe in more detail the
organization and the capabilities of the software. Finally, HEVs can be classified into two main architectures, as
using the description of the mechatronics approach, we shown in Figure 1 (mechanical or electrical power
explain the HIL methodology used to validate and

Thermal power Electrical power

HEVs

Mechanical power addition Electrical power addition

Series
Parallel (torque addition) Power split (speed addition)

Single shaft Double shaft

improve our models. Figure 1: Classification of HEVs
addition). For a parallel or power split configuration where k is a constant parameter.
(mechanical addition), the power is provided in a
mechanical way (both the electrical and mechanical In the double-shaft configuration, both of the speeds are
sources can be used directly at the wheel to propel the independent (k is now a variable and not a fixed
vehicle). In the series configuration (electrical addition), parameter anymore).
the power supplied is electric (electric motors provide all
the energy used at the wheels). PSAT allows the simulation of more than 20 parallel
configurations, including:
Series Configuration
• Three possible transmissions,
PSAT allows the user to choose, as the source of • Four positions to add the torque of the motor, and
thermal power, between an engine and a fuel cell. More • The possibility to use a fixed ratio in between the
than 20 configurations are thus available: motor and the main shaft.

• Three possible transmissions (manual, automatic, The user can choose where to place the electrical motor.
and CVT), Four positions are proposed, as shown in Figure 2:
• Two different thermal sources, and
• Several different axle ratios (e.g., none, fixed gear). • Pos1: between the engine and the link element
(clutch/torque converter),
Parallel Configuration • Pos2: between the link element and the trans-
mission,
When we consider the parallel configuration, we can still • Pos3: between the transmission and the final drive,
divide it into two main architectures [Rimaux, et al., and
1998]: • Pos4: between the final drive and the wheel.

• Single shaft and Figure 3 shows an example of a parallel hybrid.

• Double shaft.
Power Split Configuration
In the single-shaft configuration, the rotational speeds of
the engine and the motor are linked by a fixed A special case of the double-shaft hybrid configuration is
proportional ratio as follows: the power split hybrid, which use a planetary gear, an
engine, and two motors. One example of this
WICE = k * WEM configuration is the Toyota Prius, as shown in figure 4.

1 2 3 4

Figure 2: Four Positions of The Parallel Conficuration

 Figure 3: Example of a Parallel Hybrid Configuration in Position 2

Figure 4: Schematic of Prius Planetary Gear Layout

PSAT ORGANIZATION AND CAPABILITIES The output ports are used in the power controller (PTC)
for post-treatment, plotting or information in the strategy.
MODEL STRUCTURE The second ports carry the effort (i.e., voltage, torque)
and the last ones the flow (i.e., current, speed).
Organizational Format
To have an optimal reusability of the models, each
In order to exchange easily the models and implement component has its own:
new ones, we use a common format between the
input/output of the power ports by using the Bond Graph • Initialization file,
philosophy, as shown in Figure 5. The first ports are • Scaling file,
used for the information: • Model,
• List of variables (output information), and
• Input: on/off engine, gear number, etc. and • Selector file.
• Output (sensors): torque, rotational speed, current,
voltage, etc.
 Command from Info to PTC
PTC

MODEL
Effort Effort
Flow Flow

Effort = Torque, Voltage…

Flow = Speed, Current…

Figure 5: Global Formalism for the I/O of the Models Using Bond Graph

ORGANIZATIONAL STRUCTURE parameters and variables by using the Tags), the name
of the Tags are defined as follows:
Use of Library
From_’variable name”
To ensure that the models we are using are the last Goto_”variable name’
ones changed or are not modified, we decided to use a
library in which all the models are saved. Libraries Use of Selectors
enable users to copy blocks into their models from
external libraries and automatically update the copied The Selector file allows us to parameterize the location
blocs when the source blocks change. of each variable in an array. Indeed, to carry sufficient
information, three buses are used: one for the
Use of Masks mechanical component, one for the transmission
component, and the last one for the electrical side. Every
We masked the models in order to use the parameters bus consists of the output variables of each component.
as local instead of global variables. This approach allows To find the place of each parameter in these buses, we
us to use the same generic model twice with two use some parameterized value to establish its location.
different data sets. We use the “GOTO-FROM” method For instance, the variable “nb_fc_spd_hist” is located in
th
to carry the information in an array from the model block the 4 position of the array “nb_thermic_variables”
to the “workspace block” where they will be transformed (which consists of the information coming from the
into global variables and thus be used in post-treatment engine, the clutch, and the exhaust).
processing.
Figure 8 shows an example of the use of selectors.
Figure 6 shows an example of a mask with an electrical
motor. NOMENCLATURE OF VARIABLE NAMES

Use of GOTO-FROM Format All the names of the software have been parameterized
and follow some rules.
To simplify the model, we decided to use the GOTO-
FROM format as shown in Figure 7. As far as the At the software level where the computations are made,
models are concerned, all of the GOTO-FROM blocks everything is based on the name of the component (e.g.,
are local and located at the upper level of the model (no ‘compo’ = fc for fuel converter). In fact,
blocks are located in the subsystems). Moreover, to
facilitate the work for HIL (Control Desk access to the • The component model name is defined as
‘compo’_cm (ex: fc_cm),
 The same model can be used for
different components
Figure 6: Example of Using a Mask

Variable name

[ptc_fc_trq_hot_m in_cstr_hist] Goto_’Variable name’

Goto_ptc_fc_trq_hot_mi n_cstr_hi st

[ptc_fc_trq_hot_m i n_cstr_hi st]

From _ptc_fc_trq_hot_mi n_cstr_hist

ptc_fc_trq_hot_m in_cs tr_his t

Use of to_workspace to be
able to plot the variables

Figure 7: Example of Use of GOTO-FROM

• The initialization file is ccording to the component in which they are used and
‘compo’_init, the type of data they represent.
• The scaling file is
‘compo’_scale, POST-TREATMENT
• The calculation file is
‘compo’_calc, At the end of a simulation, PSAT will display the results
• The parameter used to choose if we scale or not corresponding to user-defined preferences (performance
gui_scale_’compo’, or energy consumption and emissions). The user can
• The parameters used to scale the component also access the information of every model by using an
gui_’compo’, easy and automatic way to plot the variables used for
• The selector file associated with the model that run. For each component, we can plot and add as
selectors_’compo’_cm, and many variables as are located in the menu.
• The name of the main library of a component
lib_’compo’. Figure 9 shows the choices available to the user after
running a simulation with a parallel configuration in
At the component level, all of the variables and position 2 with a reduction block between the motor and
parameters also follow established rules and are named the main shaft.
Mechanical bus
nb_variable

Transmission bus

Electrical bus
Figure 8: Example of Use of Selectors

Figure 9: Choices of Plot for a Parallel Drivetrain with CVT

Figure 10 shows an example of the data plot for a size of the different components. Starting with a
battery. conceptual model (used essentially for consumption), we
then improve it to a representational model (include such
NEW MODELS OR DATA IMPLEMENTATION system parameters as stiffness). Using Dspace, we
successively replace the simulated components by the
All of this structure has been developed for one goal: to real ones to finally control the whole prototype HEV
facilitate the integration of new models and data into powertrain. We then come back to the simulated models
PSAT quickly and easily. For instance, the Bond Graph as soon as a problem is noticed on the test bench to
philosophy allows us to integrate any engine model into correct it and improve the accuracy and robustness of
PSAT if the inputs and outputs are similar. Moreover, as the models.
far as the control strategy is concerned, the strategy
demands that outputs are always the same for a specific PSAT VALIDATION
configuration. For instance, in a parallel drivetrain using
a manual transmission, the outputs will be: To validate PSAT, two different approaches will be used.
To validate the models, we will use data gathered from
• Engine on/off, some existing HEV vehicles and some PNGV
• Engine torque, prototypes. This method will allow us to compare a
• Motor torque, whole drivetrain configuration. The other way to compare
• Gear number, and the simulated results with the real data is to control a
• Braking torque. partial or whole specific drivetrain on the test bench. We
have chosen a double-shaft post-transmission parallel
Another parameter that will facilitate the implementation hybrid with a CVT as our first test case. To verify each of
of new components or control strategies is the choice of our components, we decided to divide our tests into four
using only SI units. phases:

PSAT VALIDATION USING HIL • The first phase consists of validating one single
component (motor).
MECHATRONIC APPROACH • The second phase allows us to validate the engine
starting and clutching.
The mechatronic approach can be represented as • The third phase will be used to validate the CVT
shown in Figure 11. It consists of defining, according model and its ratio command.
to the customer’s expectations (e.g., acceleration, • The last phase will lead to the validation of the whole
maximum speed), the best HEV architecture and the drivetrain.

Figure 10: Example of Variable Implementation for the Battery

 Figure 11: Mechatronic Approach

During each of the phases, we will use the measured not based on what software it uses but on how the
data (e.g., torque, speed, current, voltage, transient company uses it to accelerate the development of the
next generation of fuel-efficient, environmentally friendly
emission measurements) to come back to the simulated vehicles.
models to validate them.
REFERENCES
Figure 12 shows the different steps of the HIL validation.
Combes, E., and Cottard, C. 1992. Voiture particuliere
CONCLUSION Electrique Hybride (in English), P027, Advanced Electric Drive
Systems for Buses, Vans, and passenger cars to reduce
pollution. EDS, European Association of Electric Road Vehicle
PSAT is a user-friendly simulation toolkit that allows the (A.V.E.R.E).
user to simulate a large number of different HEV
configurations and implement the commands on Rimaux, S.; Delhom M.; Combes E.; and Rault, A. 1998.
prototypes by using an HIL technique. As the Hybrid Vehicle Powertrain: Modeling and Control.
government-financed PSATs develop, our goal is to
provide a nonproprietary version to a wide range of Karnopp, D.; Margolis, D.; and Rosenberg R. 1990. System
nd
people later in 2000. Thus, by incorporating the data, Dynamics: A Unified Approach, 2 ed. John Wiley & Sons,
models, and strategies of users, we can speed up Inc., New York.
PSAT’s development for everyone’s benefit and greatly
DeCharentenay F. Delhom M. and Rault A.. 1996. Seamless
strengthen this powerful simulation tool. In this world of
Mechatronic design of an electric vehicle powertrainm
increasing global competition, the value of a company is Convergence 96 pp413-421

Matlab V5.3.1 / Simulink V3.0 Users Guide. The Mathworks,

Inc.
 Phase 1: Motor only Phase 2: Engine with clutch

Mot Dyno Eng Dyno

Torque sensor

Phase 3: Engine, clutch and CVT Phase 4: Whole drivetrain

4 Dyno
4 Dyno :
: 1
Eng Mot C
Eng C 1 V
V T
T

Figure 12: PSAT Validation Process Using HIL

arc Automotive Research Center NAC

The Reverse Engineering of a Diesel

Engine: A Unified Systems Approach
Using ADVISOR

George J. Delagrammatikas
and
Dennis N. Assanis
Automotive Research Center
The University of Michigan

ADVISOR Users’ Conference - August 24 - 25, 2000

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC
Outline
• Motivation and Background
– Objectives

• Traditional Methods of Engine Design

– ‘Library’ Approach

• Engineering Tools Utilized

– ADVISOR and Turbo-Diesel Simulation

• Application to ADVISOR
– Case Study Implementation

• Conclusions and Future Work

– Lessons Learned and Direction of Research

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

Motivation
• Federal Regulations
– Fuel economy (CAFE)
– Emissions (NOx, smog, and other pollutants)

• Public Awareness
– ‘Green’ movement
– Global warming scare

• Decrease Dependence on Foreign Oil

– Avoid another oil crisis - Energy security

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

Objectives
• Design the optimal engine for a given vehicle

• Develop a framework to design this engine

– Utilizing the proper simulation tools
– Implementing the correct optimization scheme

• Perform a realistic case study to analyze

trade-offs in design characteristics

• Extend methodology to different levels

– More vehicle systems
– Individual subsystems and components

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

Background

• Engineering tools used

– ADVISOR
» Conventional powertrain configuration selected
(Wipke, Cuddy, et al 1997)
» Feed-backward vehicle simulation

– TDES
» Turbocharged Diesel Engine Simulation
(Assanis and Heywood, 1986)
» Zero-dimensional, quasi-static and feed-forward
» Enhanced to calculate an engine map for ADVISOR ‘on-the-fly’

– MATLAB
» SIMULINK - ADVISOR calculations
» Optimization framework - Sequential Quadratic Programming (SQP)

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

‘Library’ Approach
ADVISOR ENGINE LIBRARY

RUN ADVISOR

CHOOSE A DIFFERENT
OUTPUT METRICS ENGINE TO ACHIEVE
RESULT DESIRED OUTPUTS

DOES ENGINE
PRODUCE DESIRED NO
VEHICLE OUTPUTS?

YES

END

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

High-Fidelity ‘Library’ Approach

OPERATING
GEOMETRIES CONTROLS CONDITIONS

ENGINE
SIMULATION
CREATE MAPS ‘ON-THE-FLY’ CHOOSE DIFFERENT
ENGINE
CHARACTERISITCS
RUN ADVISOR TO ACHIEVE
DESIRED OUTPUTS
OUTPUT METRICS
RESULT

DOES ENGINE
PRODUCE DESIRED NO
VEHICLE OUTPUTS?

YES BORE, STROKE, COMPRESSION RATIO,

VALVE SIZES AND TIMINGS, TURBO
END SETTINGS, TINJ, MANIFOLDS, ETC.

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

Issues with ‘Library’ Approaches

• Maps on file are for a given engine, under
given operating conditions
– Engine characteristics can vary dramatically
» Different auxiliary devices used (turbo, supercharger, intercooler, etc.)
» Different control strategies (VVT, VCR, VDE, etc.)

• Maps ‘on-the-fly’ can be made quite rapidly

– Changing parameters within the engine cannot be
detected by an optimizer
» Must account for noise level of both simulations

– ‘Mix-and-match’ techniques are time-consuming

» Do not afford the user with great certainty that the final result is the
true optimum

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

Model Decomposition
• Breaking up the large scale system into more
manageable subsystems
– Ensure conceptual simplicity of overall system
– Decrease computational expense
– Provide a methodical manner to perform trade-off
analyses

• Modularity of base simulations used helps

determine the decomposition method
– ‘Natural’ break-up in this case (vehicle
components)

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

Decomposition and Coordination

1 - ADVISOR Subproblem
VEHICLE
PARAMETERS 2 - TDES Subproblem

3 - Final Execution

1 ADVISOR
3 ADVISOR

2
BSFC TABLE REAL MPG,
MAX TRQ REAL PERFORMANCE

USE POINTS OPT MAX TRQ

CURVE TDES

ENGINE
PARAMETERS

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

Case Study Description

• Pre-optimality
– Find out which parameters are important
– Determine step sizes for optimization
– Investigate different objectives to optimize

• Optimization
– Execute the decomposed, coordinated
subproblems and link their results

• Post-optimality
– Perform trade-off studies to understand the impact
of different parameters on the location of the
optimum design

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC
Parameterized Torque Curves
250

200

PERFORMANCE
TORQUE

150 RUN

100

FUDS
50
CYCLE
0
0 500 1000 1500 2000 2500 3000 3500 4000 4500

SPEED
250
250

200
200

TORQUE
TORQUE

150
150

100
100

50
50

0
0
0 500 1000 1500 2000 2500 3000 3500 4000 4500
0 500 1000 1500 2000 2500 3000 3500 4000 4500

SPEED SPEED
REVERSE ENGINEERING A DIESEL ENGINE
arc Automotive Research Center NAC

ADVISOR Subproblem: Preoptimality I

MIN
RATED
TORQUE
B
TRQ2
C
TRQ3 MIN
RATED
POWE
TRANS
R
PERF

TRQ4 D

TRQ1
CONSTANT
A
CITY HWAY POWER

4
1

PM

PM
PM

PM

R
R

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

ADVISOR Subproblem: Preoptimality II

TRQ2

TRQ3

1
TRQ4

4 2
TRQ1

5
1

4
PM

PM

PM

PM
R

R
REVERSE ENGINEERING A DIESEL ENGINE
 arc Automotive Research Center NAC

Engine Matching Scenarios

TORQUE
TORQUE

CAN LIMIT FUELING CANNOT SUPPLY

STRATEGY TO OBTAIN PERFORMANCE
IDEAL TORQUE CURVE DEMANDS

RPM RPM

CORRECT INCORRECT

REVERSE ENGINEERING A DIESEL ENGINE

 arc Automotive Research Center NAC

TDES Subproblem: Engine Matching

HISTORY
TORQUE

FINAL RESULT

TORQUE
SPEED

SPEED
REVERSE ENGINEERING A DIESEL ENGINE
arc Automotive Research Center NAC
ADVISOR Subproblem:
Problem Statement
minimize (Rated Torque + Rated Power) =

f(RPM1, RPM2, RPM3, RPM4,

TRQ1, TRQ2, TRQ3, TRQ4)

subject to:

0 – 60 mph 9 12 sec.
40 – 60 mph 9 5.3 sec.
Maximum speed , 100 mph
0 - 85 mph 9 24 sec.
5 sec. Distance , 140 ft.
max acceleration , 0.5 g’s
cruising grade (55 mph) , 6.5%
max launch grade , 30%
max difference between vehicle and driving cycle speed
traces 9 2 mph
number of gear shifts 9 180
TRQ3 9 TRQ2
TRQ4 9 0.7 (TRQ2)
TRQ1 9 0.5 (TRQ2)

REVERSE ENGINEERING A DIESEL ENGINE

 arc Automotive Research Center NAC

ADVISOR Subproblem: Variable Ranges

TRQ2
TRQ3 VARIABLE LOWER BOUND UPPER BOUND
RPM1 750 850
RPM2 1800 2500
RPM3 2550 3000
RPM4 4350 4450
TRQ1 50 150
TRQ2 70 250
TRQ4 TRQ3 70 250
TRQ4 50 190

TRQ1
2
1

4
PM
PM

PM

PM
R
R

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

TDES Subproblem:
Problem Statement

minimize f = engine displacement

subject to:

ADVISOR TRQ(i) 9 TDES TRQ(i)

where i = 1 to 16

Variables:

DISP: engine displacement

CMRTIO: compression ratio
PIM: inlet manifold pressure at wastegate activation point
WRPM: wastegate activation point (RPM)
PIMMAX: maximum inlet manifold pressure

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

TDES Subproblem: Engine Variables

PIMMAX
2.5
PINLET

2
PIM
1.5

0.5

0
200 5000
150
100
2000
3000
4000
WRPM
50 1000

TORQUE 0 0
RPM

2.5

PEXHAUST 1.5

ENGINE DISPLACEMENT 1

COMPRESSION RATIO 0.5

0
WRPM 200 5000
4000
PIM 150

100
3000
2000
PIMMAX 50 1000
RPM
TORQUE 0 0

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

Final Maximum Torque Curve

BASELINE
220

TDES
200
OPTIMUM

180
TORQUE

160

140

120 ADVISOR
OPTIMUM
100
1000 1500 2000 2500 3000 3500 4000 4500

SPEED

REVERSE ENGINEERING A DIESEL ENGINE

 arc Automotive Research Center NAC

Baseline vs. Optimum Engine Map

220

200
231
BASELINE
180

246 277
TORQUE

160 262

140

120
OPTIMUM
100 324

80
356

60 293
434
309 371
402387 465 180
40 340 418 480
512
527 558605 652
449 496 543
574 683
590 621 730 824 715
699 761
777
793 746 808 933
839
637
20 668 855 871949886
902 980918 964 160
1000 1500 2000 2500 3000 3500 4000 262
140

SPEED

TORQUE
120 231

100 246
309

5.6% Better Fuel Economy 80

340

in the FTP Cycle - almost 60

277
293
371
387
418

identical performance to 40
402
434
324
356
465
496
512 543
449
480
527558 590
621
637
668 699
574
605 746 808 855
777
730 761
683652 886
baseline 20

1000 1500 2000

715
2500 3000
793
839 871 902 933 918
824

3500 4000
949
96
9

SPEED
REVERSE ENGINEERING A DIESEL ENGINE
 arc Automotive Research Center NAC

Post-Optimality: Cargo Mass Study

220
200
TORQUE (N-m)

180 136 kg
160 200 kg
140 250 kg
120 300 kg
100
80
600 1600 2600 3600 4600
ENGINE SPEED (RPM)
70
POWER (kW) 60
50 136 kg
40 200 kg
30 250 kg
20 300 kg
10
0
600 1600 2600 3600 4600
ENGINE SPEED (RPM)

REVERSE ENGINEERING A DIESEL ENGINE

arc Automotive Research Center NAC

Research Direction
• Emissions modelling
– Zero-, quasi - dimensional, and multi-zone models

• Transmission parameters
– Gear ratios and upshift/downshift strategies

• Trade-off analyses
– Vehicle mass, driving cycles, and product platform
design

• Methodology refinement
– Different parameterization techniques as well as
coordination methods

• HEV system applications

REVERSE ENGINEERING A DIESEL ENGINE
arc Automotive Research Center NAC

Possible Decomposition of HEV System

VEHICLE ENGINE TORQUE MOTOR TORQUE BATTERY TRANSMISSION
PARAMETERS CURVE CURVE PARAMETERS PARAMETERS

ADVISOR

ENGINE USE POINTS MOTOR USE POINTS BATTERY GEAR HISTORY

SOC HISTORY
+ + +
+
OPTIMAL TORQUE OPTIMAL TORQUE OPTIMAL
CURVE CURVE OPTIMAL TRANSMISSION
BATTERY PARAMETERS
PARAMETERS

ENGINE MOTOR BATTERY TRANSMISSION

SUBPROBLEM SUBPROBLEM SUBPROBLEM SUBPROBLEM

MOTOR BATTERY TRANSMISSION

TDES MODEL MODEL MODEL

REVERSE ENGINEERING A DIESEL ENGINE

 The Reverse Engineering of a Diesel Engine:
A Unified Systems Approach Using ADVISOR
George J. Delagrammatikas and Dennis N. Assanis
Automotive Research Center
The University of Michigan
Ann Arbor, MI

ABSTRACT often contain simplified models of vehicle components

that should be replaced or enhanced by a researcher on
The need for a rigorous systems engineering approach a need-driven basis.
to automotive powertrains has been addressed in this
work from the perspective of the diesel engine. A high- In order to understand the effects of powertrain design
fidelity engine simulation has been integrated with a total on overall vehicle system behavior, we must first
vehicle model for the purpose of reverse engineering the describe the vehicle analytically and perform meticulous
optimal powerplant for a given vehicle mission. Engine studies on its feasible design domain. Therefore,
parameters have been coordinated between the validated simulations must be implemented within design
simulations to develop a framework for total vehicle frameworks to allow for more rigorous examinations of
design. The design strategies discussed in this paper vehicle design. For this purpose, the ADvanced VehIcle
allow engine researchers to set targets for individual SimulatOR (ADVISOR), a public domain software
system components and to analyze the tradeoffs package, was developed by the National Renewable
associated with different vehicle mission objectives. A Energy Laboratory to aid in the analysis of Hybrid
detailed case study employing these techniques is Electric Vehicles (HEV’s), electric, and conventional
presented for a conventional vehicle where the most (engine-only) systems.
fuel-efficient engine is found that simultaneously
conforms to the desired performance criteria. The diesel engine is described within ADVISOR by an
associated maximum torque curve, a brake specific fuel
INTRODUCTION consumption (BSFC) table, and, if available, emissions
maps. These graphs are available through steady-state
The increasingly stringent requirements on fuel-efficient tests on an experimental engine. Because the data for
and environmentally-friendly automotive primemovers different engines are found within ADVISOR’s library of
has encouraged the development of programs, both in components, the researcher’s complete system designs
industry and academia, to promote the research of are limited to the number of engines available in that
alternative powerplants. The Partnership for the Next library and by the operating control strategies employed
Generation of Vehicles (PNGV), a consortium of by those engines during those tests. That is, for
automobile companies, research laboratories, and geometrically identical engines, the control strategies
universities, is involved in the discovery and used in the steady-state tests unnecessarily constrain
development of new technologies intended to provide the candidate engine designs during simulation studies.
the motive force for vehicles of the future. More A higher-fidelity and, therefore, more versatile engine
specifically, the advancement of alternative fuels and simulation is needed in order to overcome these
hybrid powertrain technologies is currently the primary artificially-imposed limits on the design domain.
PNGV goal.
One approach for predictive engine design would be to
In order to achieve this objective, the industrial, model engine outputs as functions of bore, stroke,
governmental, and institutional researchers have speed, load, and equivalence ratio within large scale
conducted field-testing and computational studies for the vehicle systems. The correlations found in these
purpose of screening candidate technologies. Though functions are very helpful during the preliminary design
simulation work is associated with great time and cost process, but the engine, as a subsystem, must be
savings for the designer, the implemented models must modeled in a manner that would allow for component
be validated through prototype verification. Overall level design. For this reason, higher-fidelity simulations
system simulation and design is an iterative process have been developed which implement first principles in
which rests heavily on the coordination of data between their predictive capabilities. Though more time-
the modeler and the experimenter. Such overall system consuming to validate and execute, these types of
simulations exist for various vehicle configurations and simulations allow the user to vary lower-level quantities,
range from proprietary automobile company codes to such as engine geometries and thermodynamic
easily downloadable shareware. These simulations properties. In both these approaches, the lower- and
higher-fidelity submodels can be integrated and drivetrain components. ADVISOR offers the designer
executed simultaneously (as in ADVISOR). More great flexibility in changing many of the models found in
specifically, the entire system model is a transfer it. These models act as placeholders that can be
function for all system inputs (regardless as to their replaced by higher fidelity models when detailed studies
effects on vehicle metrics) and outputs. Engine design regarding a particular subsystem are required.
using this ‘all-at-once’, or forward problem solving
technique though, is iterative and computationally Each block represents a component of the calculation
intensive for either engine modeling approach. that determines vehicle fuel economy and performance
metrics for a specified driving cycle. The block diagram
A more relevant and challenging exercise in the system starts on the far left with data regarding the actual cycle
design process is to identify the optimal engine for the through which the vehicle is to be driven. Next, vehicle
vehicle’s mission. This process, which is the focus of velocity is passed to a load-calculating block that finds
this work, is termed ‘reverse engineering.’ Here, the the total load on the vehicle (including inertial,
system (vehicle) level objectives are converted into aerodynamic and rolling resistance). Then, the
subsystem (engine) level targets which are then used to proceeding blocks calculate the loads and speeds that
match a realistic subsystem to attain the vehicle’s the engine and/or motor must output in order to
mission. The design of the engine’s subsystems and accelerate the vehicle to the required vehicle speed.
components must be coupled to a vehicle model for
overall system design and the data between different The ADVISOR simulation style is called feed-backward
models must be coordinated in a systematic fashion. since the flow of control begins with the torque required
This process insures minimal computational expense at the tire and ends at the fuel flow rate required by the
and greater certainty that the final result is the true engine. In real life, a vehicle operator has control of the
optimal design. fuel pedal and varies its position in order to get the
required torque to achieve a desired speed. The first
A procedure to allow the powertrain designer to quantify challenge in this research was to integrate ADVISOR
the tradeoffs associated with engine design at the with a high-fidelity engine simulation that is feed-forward
component level is illustrated in this paper. The in nature.
following work describes a methodology for powertrain
optimization using enhanced versions of ADVISOR and DIESEL ENGINE SIMULATION
a turbocharged diesel engine model. A technique has
been developed to define the engine torque The Turbocharged Diesel Engine Simulation (TDES)
characteristics required to accomplish a specific vehicle used in this work is a modified version of the parent code
mission. Then, an engine simulation has been first developed by Assanis and Heywood [1986]. TDES
implemented to find the most fuel-efficient engine that is a zero-dimensional, quasi-static, feed-forward engine
could deliver the performance requirements of that simulation which predicts engine outputs at a single
vehicle. In effect, this procedure solves the inverse operating point (engine speed and fueling rate
problem – the vehicle determines the properties of the combination). The diesel four-stroke cycle is treated as a
optimal powerplant it requires. sequence of continuous processes: intake, compression,
combustion (including expansion), and exhaust. Quasi-
Implementation of a high-fidelity engine model with steady, adiabatic, one-dimensional flow equations are
ADVISOR will be discussed along with the versatility in used to predict mass flows past the intake and exhaust
design achieved from their interaction. Then, the valves. Combustion is modeled as a uniformly
fundamental design issue posed in this research will be distributed heat release process, using Watson’s
developed and extended to the simulations used. correlation [Watson 1980]. Convective heat transfer in
Finally, a representative case study to illustrate the the combustion chamber is modeled using a Nusselt
flexibility of this design framework is presented and number correlation based on turbulent flow in pipes and
discussed. the characteristic velocity concept [Assanis and
Heywood 1986] for evaluating the turbulent Reynolds
ADVISOR: A BRIEF DESCRIPTION number in the cylinder. The characteristic velocity and
length scales required by these correlations are obtained
ADVISOR is a MATLAB/SIMULINK-based, feed- from an “energy cascade” zero-dimensional turbulence
backward simulation for HEV and conventional model [Tennekes 1972]. Radiative heat transfer is
powertrains [Wipke and Cuddy 1997]. ADVISOR allows added during combustion [Heywood 1988]. The
for analysis of the performance, emissions, and fuel combustion chamber surface temperatures of the piston,
economy of conventional, electric, and hybrid vehicles., cylinder head, and liner can be either specified or
and has been validated on numerous occasions in the calculated from a specification of the wall structure. A
open literature [Senger 1997]. The component models friction sub-model based on the Millington’s and Hartles’
in ADVISOR are empirical, relying on input/output correlation [Millington and Hartles 1968] is used to
relations measured in the laboratory, and quasi-static, predict the engine friction losses and convert indicated to
using data collected in steady state tests and correcting brake quantities.
them for transient effects, such as the rotational inertia of
The calculations above are made for an individual (WRPM) have been added to this simulation in order to
cylinder only (termed the master cylinder) and are then simulate a realistic engine of this type. The turbine and
imaged for the number of cylinders specified by the user. compressor efficiencies were held constant throughout
The interaction between the master cylinder model and the engine speed and load range while exhaust
the other components is accounted for in the manifolds backpressures were predicted. Example pressures are
modeled as separate control volumes. There is illustrated in Figure 1. By parameterizing the PIM and
instantaneous mixing of all mass flows that enter the WRPM, TDES now enables the user to perform studies
intake manifold with the gases in the manifold. To regarding novel inlet boost techniques including VNT,
complete the system, a compressor is connected to the multi-stage turbochargers, and supercharger design.
inlet side of the intake manifold and a turbine is These investigations include the reverse engineering of
connected to the outlet side of the exhaust manifold. auxiliary pressure boost designs in the same manner
The fundamental equations for turbocharger design can that the overall engine subsystem is designed in the
be found in [Watson and Janota 1982]. work presented herein, and are currently being
researched by the authors.
PIMMAX
SOLVING THE FORWARD PROBLEM: USING
2.5
DOCUMENTED ENGINE DATA

2
An ADVISOR user investigating the effects of engine
INLET PRESS

1.5 parameters on overall vehicle metrics would typically

PIM select a suitable engine, transmission, and chassis for a
1
particular application. Once ADVISOR is invoked, the
0.5 performance metrics and fuel consumption criteria are
0 WRPM calculated. At this point, the user would either vary the
200 5000 engine altogether or scale it linearly (using the
150 4000

100
3000 associated torque and speed scale parameters). This
2000
50 1000
procedure, illustrated in Figure 2 (a), is time-consuming
TORQUE
0 0
RPM and does not assure the user that the resulting engine
configuration is optimal, or even realistic. To guarantee
that the engine is a plausible design, we must introduce
(a) TDES as the engine model and link it, either directly or
indirectly, with ADVISOR.

The motive behind linking TDES and ADVISOR was to

separate all the engine-related variables and parameters
from those that describe the rest of the overall vehicle.
2.5 In so doing, more accurate studies regarding the engine
EXHAUST PRESS

2
subsystem could be performed because the overall
system simulation’s noise level would not adversely
1.5
affect the TDES calculations. The flow of this process,
1 illustrated in Figure 2 (b), is contrasted with the
0.5 traditional approach mentioned earlier in Figure 2 (a).
0
TDES outputs though, may still be masked by underlying
200 5000 vehicle simulation results if it were embedded within
4000
150
3000
ADVISOR as in [Fellini et al 1999][Sasena et al 1999].
100
2000 The following example illustrates this point.
50 1000
0 0
TORQUE RPM Let us consider that the engine designer would like to
understand the effects of varying bore size on the overall
fuel economy of a vehicle driven through the Federal
(b) Urban Driving Schedule (FUDS). Before TDES and
Figure 1: TDES parameterized inputs for inlet manifold ADVISOR were coupled, this type of study was not
pressures (a) and their associated, predicted exhaust possible. Now that the bore can be varied using TDES,
manifold pressures (b). an accurate prediction of engine characteristics can be
produced for an engine by only varying the bore size. A
For this study, TDES was developed to emulate variable new engine map could be produced ‘on-the-fly’ and sent
geometry turbocharging (VNT) with a wastegate. The directly to ADVISOR from TDES for system calculations.
parameters that describe the inlet manifold pressure
(PIM), the maximum boost pressure (PIMMAX), and the
engine speed at which the wastegate becomes activated
 models while still keeping the information pertinent to
ADVISOR ENGINE LIBRARY their individual calculations consistent. The techniques
illustrated in the following sections will address these
issues.

SOLVING THE INVERSE PROBLEM: CREATING

OPTIMAL ENGINE DATA
RUN ADVISOR
CHOOSE A DIFFERENT Among the various system outputs that are crucial in
OUTPUT METRICS ENGINE TO ACHIEVE vehicle design are fuel economy, emissions, and
RESULT DESIRED OUTPUTS performance; the federal regulations and PNGV
constraints for which will be detailed shortly. Of the
DOES ENGINE three, we will now assume that the principal factor
PRODUCE DESIRED NO
behind the acceptance of a vehicle in a specific market
VEHICLE OUTPUTS? is performance. Regardless as to how clean and
YES
efficient an engine is, it cannot be an admissible design
for a vehicle if it does not supply the adequate power
and torque levels required by the drivetrain and the
END
consumer. Though extremely important design
quantities, fuel efficiency and emissions characteristics
under different speeds and loads must be optimized for
(a) the engine that already meets the performance criteria.
To replicate the design process in the laboratory, we will
shortly illustrate one powertrain design iteration through
simulation and extend this procedure with an
OPERATING
optimization framework.
GEOMETRIES CONTROLS
CONDITIONS
Breaking down an overall system model into smaller
ENGINE
submodels is performed in a methodical manner in order
SIMULATION to decrease computational expense and to ensure
CREATE MAPS ‘ON-THE-FLY’ CHOOSE DIFFERENT conceptual simplicity. This procedure generally allows
ENGINE greater modularity among the many submodels involved,
CHARACTERISITCS
RUN ADVISOR thus increasing the potential for their parallelization on
TO ACHIEVE
DESIRED OUTPUTS different computers and/or optimization routines
OUTPUT METRICS
[Michelena et al 1999][Scheffer 1997]. Detailed
RESULT
descriptions of the basic groups of decomposition
DOES ENGINE strategies can be found in the open literature
PRODUCE DESIRED NO [Papalambros et al 1997].
VEHICLE OUTPUTS?

YES
Depending on the modularity of the models found in the
large-scale simulations being investigated, the user will
find that one decomposition method may be more
END
attractive to implement than others. In this study, the
natural break-up of the subsystems, or object
(b) decomposition, was most suitable.

Figure 2: Typical engine design method using ADVISOR Figure 3 shows how ADVISOR and TDES were
currently (a). Technique implemented with an integrated separated and what variables linked the two simulations.
engine model like TDES (b). Since the programming styles of the two simulations
(feed-backward and feed-forward, respectively) did not
However, because the approximations present in allow the two simulations to communicate, a modified
ADVISOR’s correlations bring with them inaccuracies to Newton-Raphson techinque was developed to deliver an
the final output, how is the designer to determine the engine map automatically from TDES to ADVISOR in the
true effect of a slight change in bore to the output MPG? required data format [ARC98-1][ARC98-2].
Similarly, if these simulations were placed within a
gradient-based optimization routine, how could the Block 1 and 2 represent the two decomposed
optimizer converge properly when the derivatives found optimization subproblems (ADVISOR and TDES,
at each design point were affected by the variable respectively). In Block 1, ADVISOR interprets the
change and by an unknown noise or inaccuracy level in performance constraints placed on the vehicle as a
the simulation? There is now a need to separate these maximum torque profile required by the engine. Upon
finding the optimum curve to satisfy the vehicle mission, that is achievable by a real engine requires interaction
TDES matches the torque curve in Block 2. After finding with TDES.
the most fuel-efficient engine to do so, TDES then
calculates the entire BSFC map and supplies Block 3 250
with the data. In the final block, ADVISOR simply PERF
calculates the true fuel consumption for the cycle using
200
the data from TDES.
150
1 - ADVISOR Subproblem
VEHICLE
PARAMETERS 2 - TDES Subproblem 100

3 - Final Execution
50
1 ADVISOR
3 ADVISOR

0 FUDS
2 0 500 1000 1500 2000 2500 3000 3500 4000 4500
BSFC TABLE REAL MPG,
MAX TRQ REAL PERFORMANCE

USE POINTS OPT MAX TRQ

CURVE TDES

ENGINE 250 PERF

T O R Q U E S (N-m)
PARAMETERS

200

Figure 3: Schematic of decomposition procedure applied to

ADVISOR and TDES.
150

100

Table 1: PNGV performance constraints used to define the

engine’s minimum required torque envelope. 50

0 – 60 mph ≤ 12 sec. 0
0 500 1000 1500 2000 2500 3000 3500 4000 4500
FUDS
40 – 60 mph ≤ 5.3 sec.
Maximum speed ≥ 100 mph
0 - 85 mph ≤ 24 sec.
5 sec. Distance ≥ 140 ft.
max acceleration ≥ 0.5 g’s 250 PERF
cruising grade (55 mph) ≥ 6.5 %
max launch grade ≥ 30%
200

Varying the Maximum Torque Curve – Because a real 150

engine does not yet exist for our purposes and, realizing
that the vehicle must have adequate acceleration to be 100
acceptable, we parameterize the maximum torque curve
of that engine. This curve then replaces the data for the
50
engines previously used within ADVISOR and the
associated BSFC and emissions calculations are
0
suppressed. Thus, we can convert the parameters that 0 500 1000 1500 2000 2500 3000 3500 4000 4500 FUDS
define the engine’s torque curve into the variables used
in parametric studies or in an optimization routine. ENGINE SPEED

The torque curve can be parameterized in a variety of

methods that should be determined by the designer of Figure 4: Three arbitrary, parameterized torque curves
the overall system. In this study, we have chosen to that were used in ADVISOR.
describe the maximum torque curve in a piecewise linear
fashion. Three parameterized curves are shown in ENGINE MATCHING – The above subsections aimed at
finding the profile of a maximum torque curve that would
Figure 4. Note that the (+) symbols are actual engine use meet the vehicle’s performance. The procedure
locations during the FUDS cycle and the (o) symbols are described allows the designer to vary the location of the
for the performance run. Note that all of these curves are engine use points under the maximum torque curve.
arbitrary and that developing a legitimate torque profile
How does one now reverse engineer the engine that will CASE STUDY OVERVIEW
fit these torque demands while simultaneously meeting
the ever-stringent fuel economy regulations? Also, how Now that the simulations have been coordinated in the
can the linking between the simulations be developed above manner and the required torque curve
such that the transmission, torque profile, emissions, characteristics can be varied, we can formulate the
and BSFC characteristics of the powertrain are problem which best describes the vehicle’s mission.
optimized at the system level?
A small passenger car’s characteristics were chosen
TDES Torque Curve Matching Method – Because TDES from the ADVISOR library of vehicle properties. The
was enhanced to automatically produce a complete overall objective of this vehicle was to minimize fuel
maximum torque curve and BSFC map, it can now be consumption during the Federal Test Procedure, which
wrapped within an optimization framework where it incorporates urban and highway driving patterns, as well
attempts to find a real engine which comes as close as as adhere to a number of minimum performance metrics.
possible to supplying the desired performance from These performance requirements were listed in Table 1.
ADVISOR, but can still remain as fuel-efficient as
possible. The goal here is to guarantee that the real The design procedure that will be illustrated in the
torque curve is always greater than or equal to the curve following section is composed of three distinct studies
from the ADVISOR subproblem (not that the real torque that allow the designer to perform a comprehensive
curve is as close as possible to the required curve). assessment of the resulting engine. These studies are
Figure 5 explains this fact in more detail. pre-optimality, optimization, and parametric
investigations, listed in the order in which they should be
In Figure 5 (a), the ADVISOR and TDES curves were performed. The discovery of optima through
matched in a least squares sense. We see here that mathematical means can be found in [Papalambros and
TDES cannot supply the required torque at certain levels Wilde 1991].
(regions I and II specifically). In addition, the resulting
engine has a displacement of 1.8 liters and is not highly Pre-optimality studies allow the user to determine the
boosted. Conversely, Figure 5 (b) shows a 1.7 liter that is effects of potential variables in the system to the vehicle
highly boosted and has a high compression ratio. objective and the constraints placed on that system.
Here, different vehicle parameters can be screened
according to their effect on system outputs and their
I II realistic variability within the actual system; candidate
variables can then be studied in the subsequent
optimization problem. Critical scaling factors and
convergence criteria for the optimization run are also
found in this step. These pre-optimality studies allow the
(a) (b) user to also establish different objective functions that
can be used in future trade-off analysis studies.

Upon determining the potential variables in the design,

the user must formulate the problem which best
Figure 5: Different methods of matching ADVISOR and describes the system being studied. During the
TDES maximum torque curves. (a) is correct, (b) does not optimization runs, the user may find that constraints
afford the proper vehicle performance. have to be changed to more appropriately describe
system boundaries. In addition, the objective function
Once the BSFC tables were produced with TDES for may be varied or scaled to ensure proper convergence.
both these engines and reimplemented within ADVISOR, Because Sequential Quadratic Programming (SQP) was
the smaller engine yielded a higher MPG rating. used in this case study, termination at local minima is
Therefore, the 1.7-liter would be the more appropriate expected, therefore, multiple runs at dramatically
candidate design. Choosing the proper objective in the different starting points should be employed with this
TDES subproblem is crucial in obtaining the most fuel- algorithm. This procedure is termed a ‘multi-start’
efficient engine; a fact that will be discussed shortly. technique.
The designer has the option of using either torque curve
during the final ADVISOR MPG run. Though the engine Finally, after these previous steps have been completed,
is capable of achieving higher torques, it can be parametric studies can be performed in order to better
controlled electronically not to do so through a fuel- understand the effects of parameters on the location of
limiting strategy. The differences between the fuel the optimum in the feasible design space. The designer
economies in either case will be revisited shortly. may find that the previously applied problem statement
should also be modified. ‘What if… ?’ type studies also
fall under this design step. The following section will
demonstrate these procedures.
ILLUSTRATIVE CASE STUDY it to design? In addition, in order to make fair
comparisons between engine designs, the same
Pre-optimality studies were performed on the variables technologies were used in each. Therefore, if the fuel
that describe the maximum torque curve while the consumption ratings of a 1.9l and a 1.6l were to be
transmission settings were held constant. Several compared, the optimum injection timing and realistic
important characteristics that vary the different sections implementation of inlet boosting technologies should be
of the curve were identified, along with percent changes compared. More specifically, a 1.6l engine with high
of vehicle performance results with respect to these boost and compression ratio should not be placed
changes. alongside a 1.9l naturally aspirated engine with low
compression ratio.
PRE-OPTIMALITY STUDIES - The goal of the pre-
optimality study with ADVISOR was to minimize the MIN
RATED
rated torque and power of the engine while maintaining TORQUE
the required performance demands on acceleration and B
that there were no use points that would overspeed/load TRQ2
C
the engine. In effect, this maximum torque curve would TRQ3 MIN
represent the smallest possible engine that could be RATED
POWER
placed in this vehicle. General observations made from TRANS
this study will now be discussed. PERF

Figure 6 depicts a representative maximum torque curve TRQ4 D

that was used as an input to ADVISOR to extract vehicle
performance metrics. Points A, B, C, and D are defined
TRQ1
by an engine speed (RPM) and an associated torque CONSTANT
A
(TRQ). After each execution, the engine visitation points CITY HWAY POWER
(or use points) for the city, highway, and performance
runs were plotted. The typical location and area covered

4
PM

M
by these use points are noted for each cycle.

RP

RP

RP
R
In order to attain the required acceleration demands, line
CD was constrained to be at most tangent to the Figure 6: Representative maximum torque curve and the
constant power curve of approximately 70 kW (the effects of different constraints on the heights and slopes of
vehicle that was selected needed only 66 kW to obtain its three segments.
the performance requirements). This line would be
defined by the minimum rated engine power required by Based on the claim that for the same engine use points,
the performance run. The engine’s required torque a large engine would yield higher fuel consumption than
supply was defined by the height and length of line BC. a similar quality small engine, an optimization framework
Finally, line AB was varied until a minimum required was developed to minimize both these quantities in a
distance was attained between it and the closest use multiobjective function without any weight bias. Further
point(s), that is, the points werer not allowed to exist studies concerning weights are left for future work.
above this line. For AB, the transmission parameters
(denoted by TRANS) at the low speed range were the
driving force behind this line’s slope and height. TRQ2

TRQ3
Note that many maximum torque curves were found that
could satisfy performance mathematically. Figure 7 3
shows how widely these torque curves could vary and
how unrealistic, from an engine perspective, several of
these curves were. Lines 1 and 2 are both realistic 1
curves that can be attained by the engine; we will see TRQ4
shortly which one is more desirable. Lines 3, 4, and 5 4 2
cannot be matched to an engine and are simply TRQ1
parameter combinations at local minima within the
design space. In order to understand the typical shapes 5
of real engine torque curves, TDES was then
implemented in a series of pre-optimality studies.
1

4
PM

PM

PM

PM
R

The important observations from these runs will help in

placing realistic constraints on the overall design
Figure 7: Examples of unrealistic torque curves that result
problem. A torque curve may be able to minimize the
from the ADVISOR subproblem pre-optimality study.
rated power and torque of the engine, but how feasible is
Each torque curve point in Figure 8 was allowed to move the curve realistic. In addition, both lines 1 and 2 are
within its respective lighter region, which represents the valid, but the latter does not provide the drivability
domain of the variables within the optimization problem. characteristics required by consumers. That is,
The associated ranges for these variables are listed in maximum torque is generally preferred at the low-speed
Table 2. The darker region defines the feasible design range for small vehicles. These bounds are found from
space in the problem stated in Table 3. The optimization pre-optimality studies using TDES.
was performed at several starting points; a sampling of
representative results, all of which meet or exceed the
Table 2: Range of variables for ADVISOR subproblem.
performance constraints, were shown in Figure 7. Note
that there are dramatic variations between the plots,
which indicate that the problem posed is not constrained VARIABLE LOWER BOUND UPPER BOUND
properly and/or that the final result is very sensitive to RPM1 750 850
the starting point. RPM2 1800 2500
RPM3 2550 3000
RPM4 4350 4450
TRQ1 50 150
TRQ2 70 250
TRQ3 70 250
TRQ2 TRQ4 50 190
TRQ3
The first of these realistic limits is that TRQ1 can never
be greater than approximately 55% of TRQ2, attributable
to the limits imposed by relatively low volumetric
efficiency. Parametric studies on the entire range of
engine sizes, boosting, compression, and injection
TRQ4 timing prove this fact. By the same process, TRQ4 must
be less than or equal to about 70% of TRQ2 mainly due
TRQ1 to increased friction and decreased volumetric efficiency
at high speeds. Note that these constraints only apply to
the engines incorporating the technologies and engine
qualitites noted previously. Finally, line AB in Figure 6
1

must be at least as high as the engine visitation points

PM

PM

PM

PM

through the chosen driving cycle. After further pre-

R

optimality studies, a complete optimization problem was

formulated for the ADVISOR subproblem and is detailed
Figure 8: Range of the eight variables that defined the in Table 3.
shape of the parameterized maximum torque curve in the
ADVISOR subproblem.
Table 3: ADVISOR subproblem definition.
The original constraints in Table 1 only reflected
demands from the vehicle. Though a variety of engines minimize (Rated Torque + Rated Power) =
could be matched to any one of these lines, we must first
f(RPM1, RPM2, RPM3, RPM4,
more thoroughly define the problem. Real-world
TRQ1, TRQ2, TRQ3, TRQ4)
drivability decisions were made in order to design the
engines being used in vehicles today. Among them subject to:
were achieving maximum torque at a low engine speed
and not having to shift gears extensively throughout a 0 – 60 mph ≤ 12 sec.
particular driving schedule. For these reasons, TRQ2 40 – 60 mph ≤ 5.3 sec.
must always be greater than or equal to TRQ3 (refer to Maximum speed ≥ 100 mph
Figure 6 or Figure 8 for terminology). ADVISOR was run 0 - 85 mph ≤ 24 sec.
with a baseline 1.9l engine, with data predicted by 5 sec. Distance ≥ 140 ft.
max acceleration ≥ 0.5 g’s
TDES, in order to find the baseline number of gearshifts.
cruising grade (55 mph) ≥ 6.5 %
max launch grade ≥ 30 %
Computational issues also arose during this process. max difference between vehicle
Figure 7 illustrated several unrealistic maximum torque and driving cycle speed traces ≤ 2 mph
‘optima’ which resulted during this study. Several number of gear shifts ≤ 180
obvious constraints are that TRQ2 must always be TRQ3 ≤ TRQ2
greater than or equal to TRQ3 and always greater than TRQ4 ≤ 0.7 (TRQ2)
TRQ4 (unlike the conditions seen in lines 3, 4, and 5). TRQ1 ≤ 0.5 (TRQ2)
These are mathematical constraints imposed on the
engine, but there are also engine constraints that make
Upon finding a candidate torque curve using the The next step in this process is to determine if this
ADVISOR routine, a real engine must be matched to the engine truly provides improved fuel economy at the
desired curve. For this reason, TDES receives the final required vehicle performance. The torque curve and fuel
torque curve from ADVISOR and begins to match it with map are then sent back to ADVISOR where the fuel
a viable map. Note that the curve from ADVISOR is the mileage is calculated. This calculation can be made in
minimum torque requirement for that vehicle, therefore, several ways. First, the required torque curve can be
an engine from TDES does not necessarily have to used in conjunction with the optimum TDES fuel map.
match this curve exactly Figure 5 illustrates this point. Next, the TDES torque curve can be used with the TDES
Because the torque curve from TDES is produced at an fuel map. This procedure would allow the user to
overall fuel/air equivalence ratio of 0.7, the engine’s determine whether the constant transmission settings
electronic control unit (ECU) can limit the maximum fuel can now be varied to maintain performance, but increase
allowable at each engine speed such that the theoretical cycle fuel economy as well.
curve can be met for transmission purposes. In addition,
once the engine is matched, TDES produces the BSFC
map which is later inserted in ADVISOR to perform the
necessary fuel economy calculation.

Torque (N-m)
The TDES subproblem is set up in the following manner,
shown in Table 4. Note that the different sets of curves
reflect the history of the optimization run in Figure 9 (a),
and (b) shows the final output. Let us now perform the
optimization study in the next section.

Table 4: TDES subproblem definition.

m in im iz e f = en gin e d isp lac e m en t RPM

su bje ct to: (a)

A D V IS O R T R Q (i) ≤ T D E S T R Q (i)

w h ere i = 1 to 1 6
Torque (N-m)

OPTIMIZATION STUDY – The problem statement with

constraints and variables are listed in Table 3. Note that
the ranges for the variables are the same as in Figure
10, but the feasible design space has now been properly
defined. A mulstistart routine was performed; the
optimal maximum torque curve and BSFC table of which
will be compared against a baseline engine in the
following section. Note that the curve will be as flat as RPM
possible during the midrange of speeds and is
constrained at the high speed range by the friction and (b)
volumetric efficiency limit of a real engine. The speed at
which maximum torque occurs at the low speed range Figure 9: Sample TDES subproblem history (a) and final
(line AB) is somewhat sensitive to starting position, but result (b).
because the speed difference of all the resulting outputs
is within the computational step size of that variable, the POST-OPTIMALITY STUDIES – Potential “What if… ?”
results are considered identical. This point will be type studies that can be performed in this part of the
revisited shortly and shown to be inconsequential to the investigation may include a relaxation or a tightening of
final design of the engine. the original performance metric constraints, such as
decreasing the 0-60 mph time of the vehicle or the
At this point, the torque curve was supplied to TDES allowable 5 second distance. As an example, the
where the TDES subproblem was invoked. After a authors have decided to keep the problem statements
number of multistarts, the engine with the smallest for both submodels the same but vary the cargo mass of
displacement, which matched the ADVISOR torque the overall vehicle. This study is especially of interest to
curve, was found. The resulting torque curve and the vehicle designer of small to mid-sized vehicles with
corresponding BSFC table will be contrasted with a ample passenger space or storage capacity.
baseline engine in the following section.
The optimization steps were carried out in an identical The user could now proceed to devise other similar
manner with those performed earlier except that the studies that would aid in the engine decision-making
cargo mass was varied for each case. The masses process. In addition, various driving cycles and
were ranged from 136 kg (for the baseline study) to 300 performance constraints could be placed on the vehicle
kg. The solutions to the ADVISOR subproblem are that would alter the location of the optimum engine in the
shown in Figure 10, where (a) contains the torque curves feasible design space. Let us now ascertain the validity
and (b) shows the associated power levels reached. of this methodology by comparing the resulting engine
Next, these curves were transferred to the TDES against a baseline that would have been chosen if the
subproblem whereupon the engine geometries and user implemented the traditional combinatorial technique
boost controls were matched. As one would expect, the described at the beginning of this paper.
size of the desired engine would necessarily be greater
and have moderate to high boost properties. The engine COMPARISONS WITH BASELINE
displacements for this range of engines and their
associated cargo masses are listed in Table 5. The candidate baseline engine configuration, which was
chosen as the reference for improvement, was a 1.894-
liter (nominal 1.9-liter) diesel engine with VNT, variable
220
injection timing, and high compression ratio. This engine
200
resembles the VW 1.9l TDI engine which exists in the
TORQUE (N-m)

180 136 kg ADVISOR library and is a suitable choice for the

160 200 kg selected chassis and transmission. Upon completion of
140 250 kg
the first design iteration, the resulting diesel engine was
120 300 kg
a 1.692 liter (nominal 1.7-liter) with similar inlet manifold
100 boost pressure and injection timing capabilities. The
80 maximum torque curves are shown in Figure 11 and the
600 1600 2600 3600 4600
BSFC tables are depicted in Figure 12 (a and b). Upon
ENGINE SPEED (RPM) linking these engine maps with the use points
determined by the given transmission, a fuel economy
(a) improvement of approximately 5.6% was observed for
the FTP cycle while maintaining almost identical
70 performance metrics.
60
POWER (kW)

50 136 kg
Baseline
40 200 kg
30 250 kg 220

20 300 kg TDES
10 200 Optimum
0
600 1600 2600 3600 4600
Torque (N-m)

180
ENGINE SPEED (RPM)

160
(b)

140
Figure 10: Results from post-optimality parametric study
on cargo mass. (a) shows the maximum torque curves of ADVISOR
120
the optimal engines and (b) depicts their corresponding Optimum
power curves.
100
1000 1500 2000 2500 3000 3500 4000 4500

Table 5: Optimal engine displacements for the post-

optimality parametric study. RPM

Figure 11: Comparison between baseline 1.9-liter and

CARGO MASS LITERS optimum 1.7-liter maximum torque curves.
136 1.692
200 1.831 At this point, the user has the opportunity to vary the
250 1.931 transmission parameters for further improvements in fuel
300 2.028
economy, as well as the various engine-specific
geometries and controls that would serve the similar
purpose. For instance, for the optimized engine case,
the maximum torque curve from the ADVISOR
subproblem can be used in conjunction with the
associated BSFC table for that engine. Then, the actual CONCLUSIONS
maximum torque curve is used and ADVISOR is
executed to find the new use points for that torque curve. The methodology detailed in this paper is the first
generation in the development of a framework for the
As was noted earlier, the user can now implement either preliminary design of a diesel engine within a vehicle.
the ADVISOR or the TDES curve with the output BSFC The reverse engineering practices present within this
table. When ADVISOR is run now with these quantities, framework have allowed the authors to change the
the location of the use points at the low end, defined by properties of an engine map through parametric and
low speed range of the torque curve, do not change optimization studies in order to improve the fuel
significantly and the resulting fuel consumption economy of a vehicle driven through the FTP driving
difference between approaches is negligible. One could schedule while maintaining minimum performance
now potentially investigate the fine tuning effects of the characteristics. The manner in which the system was
gear ratios and shifting strategies that would vary the decomposed and the problem statements posed for
MPG for the FTP cycle. each subproblem are specific only to ADVISOR and
TDES. Similar strategies can be applied to other
simulations if the treatment of the maximum torque curve
220
(in the system simulation) and engine parameters (in the
200 engine model) resemble that shown here.
231
180
Realistic constraints on the diesel engine’s design have
Torque (N-m)

246 262 277

160 been found through parametric studies and applied to an
140 optimization problem within ADVISOR to find the
minimum torque envelope for a given application. This
120
torque limit was then applied to an engine matching
100 324 problem within TDES in a sequential manner to find the
80
smallest, and thereby, most fuel-efficient, engine that
356
would supply the necessary power and torque
60 293
434
requirements for the specified application.
309 371
402387 465
40 340 418 480
512
527 558605 652
683
449 496
637
543
574
590 621 699 715
730 824 761
777
793 746 808 933
839 Upon matching the smallest engine for the vehicle
20 668 855 902949886
871 980918 964
1000 1500 2000 2500 3000 3500 4000
requirements, a number of variables could be changed
in order to tune the engine and the transmission to
RPM further improve fuel economy. Among these are the
BSFC contours in engine’s injection timing, manifold dimensions, and
grams/kW-hr (a)
compressions ratio as well as the gear ratios and shifting
parameters of the transmission. One must note that, if
the fuel economy is improved after each decomposition
180 step, the certainty that the final result is truly the optimal
design for the system increases, but is not guaranteed.
160
262
140 The decomposed system that has been developed in
Torque (N-m)

this framework is an outline that can be implemented for

120 231 other vehicles to design subsystems and components
individually while adhering to the overall vehicle
100 246
309
objectives. Among these systems are the various
80 configurations of the diesel-based, hybrid electric
340
vehicles as well as purely electric drivetrains which are
60
277 371
387
the focus of on-going studies by the authors.
293 418
40 324
356 449
402 465
496
512 543
480
527558 590
621
637 699
FUTURE WORK - Because this research is still in its
434 574 668
746 808 855
605
20 683652
715
730 761
793
839 824
777 886
918
871 902 933 949
96
9 infancy stage, there are many issues that have arisen
1000 1500 2000 2500 3000 3500 4000 which must be addressed in forthcoming studies. First,
the target in the ADVISOR subproblem will be
RPM
investigated from the perspective of a multiobjective
(b) function with varying weights for the rated power and
rated torque. The height of the torque plateau and the
slopes of line CD will be investigated from this viewpoint
Figure 12: BSFC tables for baseline (a) and optimal (b). to determine their effects on performance parameters.
Note that the baseline engine does have the higher peak Preliminary studies indicate that not only does the height
fuel efficiency, though this location is never used. of the curve increase, but its slopes at the low and high
speed range are also affected drastically.
Then, post-optimality parametric studies must be optimization methodologies as well as powertrain
performed to quantify the significance of different system design. Finally, Keith Wipke, Mathew Cuddy, and
values on the vehicle’s mission. That is, vehicle cargo Valerie Johnson, from the National Renewable Energy
mass, vehicle chassis weight, rolling resistance, Laboratory, assisted us in developing the tools required
aerodynamics, and transmission parameters will vary the for the application of this methodology within ADVISOR.
required engine size – but to what degree do they each
affect fuel economy? Can we use knowledge gained REFERENCES
from these studies to screen potential vehicle design
changes and determine which ones should be targeted [ARC98-1]
first for greatest fuel economy improvements? Different
driving cycle selections must be investigated along with http://arc.engin.umich.edu/arc/conf98/case3.pdf
varying the performance constraints in the ADVISOR
subproblem. [ARC98-2]

In addition, the methodology will be continued for one http://arc.engin.umich.edu/arc/conf98/hev.pdf

more fidelity step in order to determine the effects of
even smaller changes within the engine subsystem to [Assanis and Heywood 1986] D. Assanis and J.
the performance of the overall system. In this context, Heywood, “Development and Use of a Computer
efforts have already been made in the area of variable Simulation of the Turbocompounded Diesel System for
compression ratio, continuously variable transmission Engine Performance and Component Heat Transfer
capabilities, as well as flexible valve timing capabilities. Studies,” SAE Paper 860329, 1986.

The parameterization method for the maximum torque [Fellini et al 1999] Assanis, D., Delagrammatikas G.,
curve in ADVISOR will also be researched. The Fellini R., Filipi Z., Liedtke J., Michelena, N.,
sensitivity of the objective function in this subproblem Papalambros, P., Reyes, D., Rosenbaum D., Sales A.,
with respect to the ERPM points has raised several and Sasena M., "An Optimization Approach to Hybrid Electric
concerns. Fourth-order quadratic functions, which can Propulsion System Design," Journal of Mechanics of
afford rather flat plateaus and the necessary slopes in Structures and Machines, Vol. 27, No. 4, pp. 393-421,
the low and high speed ranges, are potential candidates 1999.
for this study. In addition, the PIM, PIMmax, and WRPM
parameterization scheme for the TDES subproblem will [Heywood 1988] J. Heywood, “Internal Combustion
also be modified to make the boost scenarios predictive Engine Fundamentals,” McGraw Hill, Inc., New York,
(and more realistic) with turbocharger maps or perhaps NY, 1988.
neural networks.
[Sasena et al 1999] Fellini, R., Michelena, N.,
Next, efforts will be made to incorporate an emissions Papalambros, P., and Sasena, M., "Optimal Design of
modeling and optimization strategy to the overall design Automotive Hybrid Powertrain Systems," Proceedings of the
framework. Though preliminary NOx and dry particulate EcoDesign '99: 1st International Symposium on
matter models have already been implemented within Environmentally Conscious Design and Inverse
TDES, they have not yet been validated and therefore Manufacturing, Tokyo, Japan, February, 1999.
not presented in this paper. Methods similar to those
described earlier will be implemented for this puprose. [Michelena et al 1999] N. Michelena, C. Scheffer, R.
Fellini, and P. Y. Papalambros, ‘A CORBA-based
Finally, this work is the precursor to an optimization Framework for Distributed, Object-Oriented System
strategy for the parallel HEV which will require a greater Design,’ Journal of Mechanics of Structures and
degree of coordination between submodels. Machines, Vol. 27, No. 4,1999.
Transmission and motor design and control quantities
are being screened for their potential incorporation within [Millington and Hartles 1968] B. W. Millington, E. R.
a scheme similar to the one presented above. Hartles, ‘Frictional Losses in Diesel Engines,’ SAE Paper
680590, SAE Transactions, Vol. 77, 1968.
ACKNOWLEDGEMENTS
[Papalambros et al 1997] Wagner, T., Papalambros, P.
The authors would like to thank the generous Selection Families of Optimal Engine Designs Using
contributions of the members of the Automotive Nonlinear Programming and Parametric Sensitivity
Research Center at the University of Michigan. This Analysis. SAE Paper # 971600.
research could not have been performed without the
assistance of Panos Papalambros and his group in the [Papalambros and Wilde 1991] P. Y. Papalambros, D. J.
area of design optimization techniques for large-scale Wilde, Principles of Optimal Design, Cambridge
systems. The works and advice of Nestor Michelena, University Press, New York, 1991.
Zoran Filipi, Michael Sasena, and Ryan Fellini were
influential in our understanding of the issues relating to
[Scheffer 1997] C. Scheffer, “A Software Framework for
Optimal Systems Design,” Master of Science Thesis,
Department of Mechanical Engineering and Applied
Mechanics, University of Michigan, Ann Arbor, MI, 1997.

[Senger 1997] R. Senger, “Validation of ADVISOR as a

Simulation Tool for a Series Hybrid Electric Vehicle
Using the Virginia Tech FutureCar Lumina,” Master of
Science Thesis, Department of Mechanical Engineering,
Virginia Polytechnic Institute and State University,
Blacksburg, VA, 1997.

[Tennekes 1972] M. Tennekes, J. L. Lumley, A First

Course in Turbulence, MIT Press, Cambridge,
Massachusetts, 1972.

[Watson et al 1980] N. Watson, A. D. Pilley, M.

Marzouk, “A Combustion Correlation for Diesel Engine
simulation’, SAE Paper 800029, Warrendale, 1980.

[Watson and Janota 1982] N. Watson and M. S. Janota,

“Turbocharging the Internal Combustion Engine,” John J.
Wiley and Sons, New York, NY, 1982.

[Wipke and Cuddy 1997] M. Cuddy and K. Wipke,

“Analysis of the Fuel Economy Benefit of Drivetrain
Hybridization,” SAE Paper 970289, 1997.
A Hybrid-Propulsion Powertrain with Planetary Gear Set:
Simulation Results and a Design Approach
Presentation
Paper
Marco Santoro, Dresden University of Technology
Leone Martellucci, University of Rome

Use of ADVISOR for simulation of a Hybrid Electric Vehicle

with a Stirling Engine as the Auxiliary Power Unit
Presentation
Paper
Luis Figueroa and Owen R. Fauvel, University of Calgary
Graham T. Reader, University of Windsor

GWU Hybrid Electric Sport Utility Vehicle Design

for FutureTruck 2000 Competition
Presentation
Paper
Mohd-Syaifuddin Mohd, Zeki Gokce, Vahid Motevalli, and Kartik Bulusu
Leone Martellucci Marco Santoro

A Hybrid-Propulsion Powertrain with

Planetary Gear Set: Simulation Results
and a Design Approach

University of Rome I “La Sapienza” - Italy Dresden University of Technology - Germany

• Parallel Power-Split
• Sluggish dynamic behaviour
Rear final drive

of ICE
EM

• Traction torque provided BP

INV

directly by the ICE and by MT PGS EG

VR

the electrical drive at the

Front final drive

same time
• 2 or 4-wheel-drive layouts
available
• Pure electric operation into BP: battery pack; EM: electric motor; INV: inverter;
MT: internal combustion engine; PGS: planetary gear set;
restricted zones EG: electrical generator; VR: voltage regulator
 Planetary Gear Set links
• ICE to the Sun Gear
• Generator to the Planetary Sun gear
Carrier
• Driving Shaft to the Ring to the electrica
Gear Planet gears
generator shaf

with carrier

1 τ −1
ω engine = ⋅ ω driving _ shaft + ⋅ ω generator
τ τ Internal
to the i.c. ring gear
engine shaft
ω driving_sh aft = τ ⋅ ω engine + (1 - τ ) ⋅ ω generator

τ 1
ω generator = ⋅ ω engine − ⋅ ω driving _ shaft
τ −1 τ −1
 • The electric motor driving
the rear wheels operates in
the 1st (traction motor) or
4th (regenerative braking)
quadrant
• The electrical generator
connected to the PGS
planetary carrier operates in
the 1st/2nd or 3rd/4th
M: traction torque; MR: resistance torque; quadrants, depending on
N: rotation speed
the sign of the PGS
conversion ratio τ
 gal
e x_calc
t total fuel used (gal)
Clock To Workspace AND
fc_emis

emis
HC, CO,
ve hicle controls e xha ust sys NOx, PM (g/s)
sipre hybrid
<vc> control sta te gy <ex>
ICE fuel
<cs>
conve rte r
<fc>

<sdo>

P roduct4

Ground
drive cycle ge a r ge ne ra tor/
<cyc>1 re duction controlle r <gc> powe r e ne rgy
bus <pb> stora ge <e s s>
ve hicle <ve h>1 pla ne ta ry
front whe e l a nd front
ge a r se t <pgs>
a xle <wh> final drive <fd>

veh_front_pwr_a
power available
force ava il. to P roduct2
to front axle
front whe e ls

veh_tract_pwr_a re a r whe e l a nd re a r ge a rbox <gb> motor/

a xle <wh>1 final drive <fd> controlle r <mc>
available tractive
P roduct3
power

De mux veh_rear_pwr_a
sum of
availa ble power available
P roduct
Mux tra ctive force s (N) De mux at rear axle

Force /Line a r S pe e d

4WD SIPRE 3 Simulink model

 pgs_pwr_in_a
1 Demux Mux
To Workspace6 1
torque and speed av ail. Product5
Demux Mux torque and speed av ail.
at main input of pgs
at output side of pgs
(Nm), (rad/s) (Nm), (rad/s)
---SUN GEAR--- ---PLANETARY CARRIER---
pgs_trq_in_a
(1-tau)/tau pgs_trq_out_gen_a
To Workspace2
To Workspace4
ENGINE TO GENERATOR
pgs_spd_in_a TORQUE pgs_spd_out_gen_a
To Workspace3
To Workspace5
pgs_pwr_gen_a
Block
NOTES To Workspace7
Product4
-1/tau

ENGINE TO DRIVING SHAFT

TORQUE

tau/(tau-1)

engine to generator speed

1/(tau-1)
Sum3 pgs_trq_out_a
driv ing shaf t to generator speed
To Workspace8

[N] f d_ratio_f ront |u|

Abs
From <v eh> (rad/s)
Gain
Wheel speed at pgs_spd_out_a
beginning of time
To Workspace9
step
pgs_pwr_out_a
To Workspace11
Product7

Mux 2

Mux1 torque and speed av ail.

to f ront f inal driv e
(Nm), (rad/s)
---RING GEAR---

Planetary-Gear-Set Simulink model

 Auxilec Thomson 32 kW PM traction drive

22 kW generator/controller scaled
from Mannesmann Sachs 63 kW
PM generator/controller

Electrical-machines
efficiency maps
 0.85

0.7
0.6 0.75

0.8
0.65

0.85

0.9
0.75
0.8 0.75

0.85 0.9

0.7
0.65

0.6
0.75 0.85

0.75 0.55
0.8

0.7

0.7

0.8
0.85 0.65 0.85
0.6 0.6

Electrical-machines
0.75
0.65
0.55 0.55

operating points over the

0.7 0.65 0.7
0.75 0.8
0.85 0.6
0.6
0.85

FTP cycle 0.8

0.65

0.75
0.55
 ICE Operation - Fiat - Alfa Romeo 2.4 JTD (100 kW) Turbo Diesel Engine
90

30.1
80 37.1

70 34.1

60

50
Torque (Nm)

34.1

40
36.1

30.1
30

20

22.1
26.1
10 max torque curve
design curve
output shaft
op. pts(includes inertia & accessories)
0
0 500 1000 1500 2000 2500 3000 3500 4000 4500
Speed (rpm)

Engine operating points over the FTP cycle

 MilesPG
84 NOx (grams per mile)
0.74

0.72
82
0.7
80
0.68
MilesPG

0.66
78

NOx
0.64
76
0.62

0.6
74
0.58
72
-5 -4 -3 -2 -1 0.56
-5 -4 -3 -2 -1
tau
tau

MPGGE CO (grams per mile)

74 0.1974

73

72 0.1972

71

70 0.197
MPGGE

CO

69

68 0.1968

67

66 0.1966

65

64 0.1964
-5 -4 -3 -2 -1 -5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1
tau tau
 HC (grams per mile)
PM (grams per mile)
0.018
0.0196

0.0194

0.0179

0.0192
PM

0.019

HC
0.0179

0.0188

0.0178
0.0186

0.0184
-5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 0.0178
tau -5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1
tau

Accel_0_60 (seconds)
12 Grade_55mph
8

11.5 7.5

11 7
Accel_0_60

Grade_55mph
6.5
10.5

6
10
5.5
9.5
5

9
-5 -4 -3 -2 -1 4.5
-5 -4 -3 -2 -1
tau
tau
 Kjara
• Two-seat roadster
• The engine drives
the rear wheels

• The motor drives

the front wheels
Kjara tested on
Lombardore (Turin) track.
1, Planetary Gear Set with differential;
2, Diesel engine; 3, Electrical generator;
4, Electrical motor; 5, Electronics; 6, Battery pack; 7, Diesel fuel tank.

1, Sun gear carrier; 2, External case;

3, 4, final drive with differential; 5,6, axle shaft joints

Planetary Gear Set from 1992 4x4 FORD

Escort, sold in the European market.
 A Hybrid-Propulsion Powertrain with Planetary Gear Set:
Simulation Results and a Design Approach
Marco Santoro
Dresden University of Technology - Germany

Leone Martellucci
University of Rome I "La Sapienza" - Italy

ABSTRACT of part of the ICE torque to the front-driving wheels in

order to contribute to the vehicle traction together with
A power-split hybrid powertrain adopting a planetary the full-electric rear axle.
gear set is proposed in this work as alternative to the Rear final drive
traditional hybrid-propulsion schemes.
The main advantages of this drivetrain are: EM
INV

• Sluggish dynamic behaviour of the internal BP

VR
combustion engine, like in series powertrains;
• Traction torque provided directly by the Internal MT PGS
EG

Combustion Engine (ICE) and by the electrical drive

Front final drive
at the same time, like in parallel powertrains;
• 2 or 4-wheels-drive layout;
• Pure electric operation when driving into restricted
zones.
BP: battery pack; EM: electric motor; INV: inverter; MT: internal
combustion engine; PGS: planetary gear set; EG: electrical generator;
â ã
The Simulink model which allows ADVISOR to VR: voltage regulator
generate energy flows and emissions of the vehicle is
Figure 1: SIPRE 3 in the 4WD layout
presented. Moreover, the paper explains the control
strategy and proposes a design approach of a sports-car
POWERTRAIN DESCRIPTION
adopting this powertrain.
In order to explain the PGS kinetic and dynamic
INTRODUCTION
behaviour and its influence over the entire system
operation in detail, it is necessary to illustrate the layout
The Department of Nuclear Engineering and Energy
of the various mechanical linkages. The planetary gear
Conversions at University of Rome I “La Sapienza” has
train used can be chosen between three different
studied in the last years a “family” of power-split
configurations: the Rolls Royce, the Pickering and the
drivetrains provided with a planetary gear system: the
differential one. The choice depends on the desired
SIPRE - an acronym in the Italian language: Sistema
transmission ratio, with reversal or upward conversion.
Ibrido di Propulsione con Ruotismo Epicicloidale
(Hybrid-Propulsion System with Planetary Gear Set) - RA

RA RB

family. The third member of this family, the SIPRE 3 [1], RA RB RA

O
RB
R2
B RB

O RA
is proposed in this paper. RB

O1
B
B R2

B B
O2
R2
O1
R1 R1
O2

The planetary gear set is the heart of the SIPRE 3 R1 R1 R2

R2

system. Due to the planetary-gear-set (PGS) unit it is R1

possible to share the thermal-energy driving torque Pickering Rolls-Royce Differential

between the electrical generator and the front axle; this
is the main original characteristic that distinguishes the (τ positive) (τ negative) (τ negative)
SIPRE 3 system from a traditional series-hybrid
powertrain. In fact, such a behaviour, warranted by the Figure 2: PGS layouts
presence of a PGS, allows the mechanical transmission
The main parameter that defines the PGS behaviour is with a range of speed that is a function of τ. The
its conversion ratio τ, which depends on the number of kinematics of the PGS thus supports two degrees of
teeth in the sun and ring gear, and the principal freedom, so it is possible to freely select two different
relationship among the speeds of the axes is the Willis speeds (e.g. the ICE speed and the transmission shaft
[2] one: speed).

ω2 − Ω Other power-split powertrains with planetary gear

τ= (1) systems have been proposed in the technical literature.
ω1 − Ω In the Toyota Prius [3] the ICE is linked to the planetary
carrier, the generator to the sun gear and the driving
where ω2, Ω and ω1 are the rotation speeds, respectively, shaft to the ring gear. In the drivetrain proposed by
of the internal ring gear (the transmission shaft ωdriving_shaft), researchers at Warsaw University of Technology [4] the
of the planetary carrier (the electrical generator ωgenerator) ICE is connected to the sun gear, the generator to the
and of the sun gear (the internal combustion engine ring gear and the driving shaft to the planetary carrier.
ωengine). In the Eq.1 when Ω=0 is The aim is always to select the engine’s operating point
independently from the vehicle’s speed, exploiting the
two degrees of freedom the PGS allows.
ω
τ = 2 (2)
Ω =0 ω In order to understand the system behaviour completely,
1 it is important to report the torque distribution over the
three PGS axes. In fact in the dynamic field the PGS
and τ is positive if the first and last gear rotate in the behaviour is quite different; the main relationship that
same direction (upward conversion) and is negative describes the PGS operation is:
otherwise (reversal conversion).
Tdriving _ shaft + T generator + Tengine = 0 (4)
Sun gear

where the torque notation has the same meaning as the

speed notation previously described.
to the electrica
generator shaf
Planet gears From the Willis formula and the balance of powers in the
with carrier device

Internal Tdriving _ shaft ⋅ ω driving _ shaft + Tgenerator ⋅ ω generator +

to the i.c. ring gear (5)
engine shaft
+ Tengine ⋅ ω engine = 0

it is possible to come to the following expressions:

τ
Figure 3: PGS links in the SIPRE 3
Tengine = ⋅ Tgenerator
1- τ
From the Willis formula the relationship among the (6)
various speeds can be conveniently derived as follows Tdriving_shaft 1
=-
1 τ −1 Tengine τ
ω engine = ⋅ ω driving _ shaft + ⋅ ω generator
τ τ
The dynamic problem is completely settled when just
one of the three axis torques is fixed. For an automotive
ω driving_shaft = τ ⋅ ω engine + (1 - τ ) ⋅ ω generator (3) power unit the main condition can be the driving torque
Tdriving_shaft needed for the traction, so Tgenerator and Tengine
have to exactly match the Equations 6 and the choice of
the value of the transmission ratio τ affects the kinematic
τ 1
ω generator = ⋅ ω engine − ⋅ ω driving _ shaft and dynamic behaviour of this hybrid traction system.
τ −1 τ −1
It is important to note that the vehicle's dynamic
behaviour is controlled via the brake and accelerator
pedals, acting exclusively on the electrical traction
From these relations is clear that, unlike a traditional equipment arranged on the rear axle. From this point of
generator set used in series hybrid systems, the view this vehicle is managed just like a full electric
electrical generator of SIPRE 3 works at variable speed vehicle. This is possible considering that the hybrid
equipment acting on the front wheels, which supplies an PDS = Tdriving_shaft ⋅ ω driving _ shaft =
additional traction torque to the vehicle and cooperates
with the rear electrical traction, is characterised by a fully Tdriving_shaft ω driving_shaft
automatic operation without the need of an external Tengine ⋅ ω engine ⋅ ⋅ =
management. Furthermore, the engine needs no starter
Tengine ω engine
æ 1 ö æ ω driving_shaft
because the generator can act to crank it. Because ö
there is no reverse gear, vehicle reversing is a motor- P0 ⋅ ç − ÷ ⋅ ç ÷
only drive mode. è τ ø çè ω engine ÷
ø
Looking at the relations among the speeds in the PGS, it (8)
is interesting to observe that the speed of the generator PEG = T generator ⋅ ω generator =
is zero when
Tgenerator ω generator
ω driving _ shaft = τ ⋅ ω engine Tengine ⋅ ω engine ⋅ ⋅ =
ω generator = 0
(7)
Tengine ω engine
æ 1 - τ ö æç ω generator ö
÷
and the two degrees of freedom supported by the P0 ⋅ ç ÷⋅
kinematics of the PGS allow the generator to operate in è τ ø çè ω engine ÷
ø
two different quadrants, depending on the engine’s and
nd th
vehicle’s speeds, in generating (2 and 4 quadrants) or
st rd
motoring (1 and 3 quadrants) mode (Figure 4). The
generator/controller allows thus a bi-directional flow of where PDS is the power that reaches the driving shaft, P0
power from the batteries to the PGS unit. is the power supplied by the ICE and PEG is the power
absorbed by the electrical generator.

The advantages of the PGS operation, governed by the

previous equations, are manifest when one analyses the
ratio between the power mechanically transmitted to the
front driving wheels and the total power supplied by the
ICE; in fact, this parameter provides the exact amount of
purely mechanical traction power available to the front
vehicle wheels cooperating with the rear electric motor
for the vehicle motion. In other words, thanks to the
presence of a PGS, the electric motor has to supply a
lower power than that necessary in a series hybrid drive
M: traction torque; MR: resistance torque; N: rotation speed train. Moreover, the load on the electrical generator is
lower too, with evident benefits for the overall equipment
Figure 4: The four operating quadrants of a cost.
motor/generator
A power ratio parameter can be mathematically
The electric motor driving the rear wheels operates in expressed as a function of the PGS conversion ratio τ:
st th
the 1 (traction motor) or 4 (regenerative braking)
quadrant. The electrical generator connected to the PGS PDS 1 ω driving_shaft
planetary carrier operates in the 1 /2
st nd rd
or 3 /4
th
ρ= = − ⋅ (9)
quadrants, depending on the sign of τ. P0 τ ω engine

From the previous kinetic and dynamic equations it is Assuming the speed ωengine of the ICE fixed, the ICE
easy to calculate the expressions for the power power transmitted directly to the driving shaft increases
distributions over the PGS axes (Eq. 8): linearly with the vehicle speed, related to ωdriving_shaft. At low
speed the traction power is thus mostly supplied by the
electrical drive. The PGS acts like a power-split device
and a CVT (Continuously Variable Transmission) at the
same time.

Another power ratio parameter is

PEG 1 ω driving_shaft
σ= =1- - ⋅ (10)
P0 τ ω engine
 gal
ex_calc
t total fuel used (gal)
Clock To Workspace AND
fc_emis

emis
HC, CO,
vehicle controls NOx, PM (g/s)
sipre hybrid exhaust sys
<vc> control stategy <ex>
ICE fuel
<cs>
converter
<fc>

<sdo>

Product4

Ground
drive cycle gear generator/
<cyc>1 reduction controller <gc> power energy
bus <pb> storage <ess>
vehicle <veh>1 planetary
front wheel and front
gear set <pgs>
axle <wh> final drive <fd>
#
veh_front_pwr_a
power available
force avail. to Product2 to front axle
front wheels

veh_tract_pwr_a rear wheel and rear gearbox <gb> motor/

available tractive axle <wh>1 final drive <fd> controller <mc>
Product3 power

veh_rear_pwr_a
sum of Demux
available power available
Product at rear axle
Mux tractive forces (N) Demux

Force/Linear Speed

Figure 5: Simulink SIPRE 3 HEV block diagram’s top

level.

When the vehicle is at rest (ωdriving_shaft=0), σ = 1 in Eq. 10 positive power (the electrical machine operates as
and the vehicle is a series hybrid because the whole motor) flows from the batteries to the PGS unit.
power generated by the ICE flows uniquely to the
generator. At higher speeds the powertrain tends to
behave like a parallel hybrid and increases the power SIMULATION RESULTS
flowing to the front wheels.
The vehicle used in the simulations was a PNGV-type
THE SIMULINK MODEL vehicle. Its components and characteristics included:

Figure 5 shows the Simulink model which allows • 28 kW CIDI engine (scaled from Fiat 100 kW 2.4 L
ADVISOR to generate energy flows, emissions and Turbo Diesel Common Rail Direct Injection engine)
consumption of the vehicle. The model describes the
front and rear axles of a vehicle provided with a SIPRE 3 • Auxilec Thomson 32 kW PM traction drive [6]
drivetrain in the 4WD layout and has been built
modifying ADVISOR at the block diagram level, by • 22 kW generator/controller (scaled from
programming in Simulink. Although ADVISOR relies Mannesmann Sachs 63 kW permanent magnet
heavily on a backward-facing approach for its operation generator/controller [7])
[5], it is possible to reconnect ADVISOR block diagrams
to model new vehicle types and add new component • Twenty 18 Ah lead acid batteries (Optima spiral-
block diagrams, like the PGS one, which are related to a wound VRLA, data from NREL tests)
forward-facing approach. The generator/controller block
diagram allows bi-directional power flow, depending on • Mass of 1010 kg.
the sign of power and following the convention in Figure
4: negative power (the electrical machine operates as • Coefficient of drag (Cd) of 0.3
generator) flows from the PGS unit to the batteries;
•
2
Frontal area of 2 m .
 MilesPG NOx (grams per mile)
84 0.74

0.72
82
0.7

80 0.68
MilesPG

0.66

NOx
78
0.64

76 0.62

0.6
74
0.58

72 0.56
-5 -4 -3 -2 -1 -5 -4 -3 -2 -1
tau tau

MPGGE CO (grams per mile)

74 0.1974

73

72 0.1972

71

70 0.197
MPGGE

CO

69

68 0.1968

67

66 0.1966

65

64 0.1964
-5 -4 -3 -2 -1 -5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1
tau tau

HC (grams per mile)

PM (grams per mile) 0.018
0.0196

0.0194
0.0179

0.0192
PM

HC

0.019 0.0179

0.0188

0.0178
0.0186

0.0184
-5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 0.0178
tau -5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1
tau
 Accel_0_60 (seconds) MPGGE
12 77

76
11.5

75
11
Accel_0_60

74

MPGGE
10.5
73

10
72

9.5 71

9 70
-5 -4 -3 -2 -1 0.2 0.4 0.6 0.8 1
tau tau

Grade_55mph NOx (grams per mile)

8 0.63

0.62
7.5
0.61
7
0.6
Grade_55mph

6.5 0.59
NOx

6 0.58

0.57
5.5
0.56
5
0.55

4.5 0.54
-5 -4 -3 -2 -1 0.2 0.4 0.6 0.8 1
tau tau

CO (grams per mile)

0.2135
Figg. 6-13: Parametric study. Performances over the
US FTP cycle when τ changes between
(-5) and (-1.1).
0.213

We are testing for the time being a real-time control 0.2125

strategy inspired to a “dynamic” approach recently
proposed [8]. In Figg. 6-18 we show the performance
CO

results obtained varying the PGS conversion ratio τ in a

PNGV-type vehicle controlled with a very simple “static” 0.212

power follower strategy: the optimum operational design

curve in the ICE torque/speed map is calculated by
equally weighting the importance of fuel economy and 0.2115
each emission component.

0.211
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
tau
 HC (grams per mile) aim of that project was to develop a laboratory vehicle
0.02
useful to experiment various hybrid configurations.
0.0199
The idea is to use the current layout, but adapted to a
SIPRE 3 system, maintaining a big part of the system.
0.0199 Kjara is a roadster and the powertrain characteristics,
the arrangement of mechanical components, the
0.0199 architecture of chassis and suspensions are typical of a
sports-car (see Figg.19-20).
0.0198
HC

0.0198

0.0197

0.0197

0.0196

0.0196
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
tau

PM (grams per mile)

0.0196

Figure 19: Kjara vehicle prototype.

0.0195

0.0194
PM

0.0193

0.0192

0.0191

0.019
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
tau Figure 20: Kjara vehicle prototype on Lombardore
(Turin) track.

Figg. 14-18: Parametric study. Performances over The powertrain is composed by a thermal engine and an
electric motor: the turbocharged diesel engine is
the US FTP cycle when τ changes
installed in the rear, mid-mounted, transversally. It drives
between 0.3 and 0.9.
the rear wheels through a conventional 5-speed
gearbox. The engine is a 85 kW Fiat 2.5 L turbocharged
indirect injection diesel engine. The 18.3 kW electric
motor is installed in the front, mid-mounted,
DESIGN APPROACH longitudinally, and drives the front wheels by means of a
transaxle 2-speed gearbox with differential. Its control
In order to give a detailed vision of the hybrid propulsion electronics, manufactured by Celco Profil, is put on the
system here discussed, it is useful to present the design dashboard by the right side of the steering wheel. The
approach for the realization of a working prototype. lead-acid battery pack is composed of 24 cells, each
The Department of Nuclear Engineering and Energy weighting 9.4 kg (total battery weight is 225.6 kg). Being
Conversions of University of Rome “La Sapienza” has the cells sealed, they can be put flatwise on one side
recently realized a hybrid sports-car prototype, called under the two seats.
Kjara, in collaboration with the National Research With such kind of powertrain, it is easy to design a quick
Council (CNR); Kjara [9] is a parallel hybrid car and the conversion of the system to a SIPRE 3 configuration; in
fact, with the simple and quick substitution of the Diesel The design approach here proposed can be realized in
engine currently mounted on Kjara prototype with the short time and at an affordable cost (for a University it is
planetary gear set group (see fig.21: PGS + ICE + EG + very important !) using an existent PGS. The drawing in
final drive, but on rear wheels), it is possible to switch to Fig. 22 represents in fact a feasible solution that uses a
a SIPRE 3 propulsion system. PGS with incorporated automotive differential, all
derived from 1992 4x4 FORD ESCORT. This
component, light and with a good transmission ratio,
matches well the SIPRE 3 demands, and benefits
obviously of a good reliability and quality. The adoption
of this component is a very good opportunity for a
prototype, cutting consistently the design and
construction costs.

1, Sun gear carrier; 2, External case; 3, 4, final drive with

differential; 5,6, axle shaft joints.
Figure 22: FORD PGS internal layout.

The FORD PGS internal layout, depicted in Fig.22, is

provided with the sun gear jointed to the transmission
shaft and to the differential, leaving for ICE and EG the
internal ring gear and the planet gears carrier.
Thus, considering that the SIPRE 3 layout can be
realized in any different combination of components
which can be freely mounted on the three PGS axes, as
previously explained, the solution given in Fig.22 is
characterized by the Diesel engine joined to the internal
ring gear and the electrical generator joined to the
planetary gear carrier.
In this way a full SIPRE 3 equipment can be realized, in
the four wheel drive configuration, needing only few
mechanical modifications and starting from an easy
available automotive component acting as PGS. So,
using an existing hybrid car prototype it is possible to
test on track all the simulation results ADVISOR can
generate. The aim is to put on track and test the SIPRE
3 vehicle by the end of 2000.
1, Planetary gear set with differential; 2, Diesel engine;
3, Electrical generator; 4, electrical motor; 5, electronics;
CONCLUSION
6, battery pack; 7, Diesel fuel tank.
A power-split hybrid drivetrain is analyzed in this paper.
Figure 21: Layout solution for the implementation of
Its main characteristics are:
SIPRE 3 on Kjara vehicle prototype.

In figure 21 it is possible to see the design approach • vehicle reversing is a motor-only drive mode
here proposed.
Maintaining the electric motor, its electronics and the • the generator cranks the engine
battery pack, and mounting the PGS set with a little
diesel engine and the electrical generator, it is possible • the vehicle operates electrically when driving into
in few hours of mechanical work to pass from the restricted zones
original parallel configuration to the SIPRE 3 system.
• 2WD and 4WD layouts are available
DEFINITIONS, ACRONYMS, ABBREVIATIONS
• the generator can spin backwards
ADVISOR:
ACKNOWLEDGMENTS ADvanced VehIcle SimulatOR

We wish to thank the Italian National Research Council CIDI:

(CNR) and the Italian Ministry of Universities and
Scientific and Technological Research (MURST) which Compression Ignition Direct Injection
are supporting this work.
CNR:
CONTACT Consiglio Nazionale delle Ricerche (Italian National
Research Council)
Marco Santoro holds a “laurea” degree in electronic
engineering from University of Rome I “La Sapienza”. EG:
He is currently doctorate student at University of Rome I
“La Sapienza” and guest student at Dresden University Electrical Generator
of Technology, where he is preparing his thesis about
the energy conversion efficiency into hybrid electric 4WD:
vehicles. For further information on this paper, he may
Four Wheel Drive
be contacted at [email protected].
FTP:
Leone Martellucci holds a “laurea” degree in mechanical
engineering from University of Rome I “La Sapienza” US Federal Test Procedure, “city” cycle for city-highway
and a doctorate degree from this same University. He is tests
currently researcher at University of Rome I “La
Sapienza”, Department of Nuclear Engineering and HEV:
Energy Conversions. For further information on this
paper, he may be contacted at Hybrid Electric Vehicle
[email protected].
ICE:
REFERENCES Internal Combustion Engine

1. Landolfi O., Martellucci L., "New propulsion system for MURST:

th
hybrid vehicle", SAE Paper 931882, 7 International
Pacific Conference, November 1993, Phoenix, Arizona Ministero dell'Università e della Ricerca Scientifica e
(USA). Tecnologica (Italian Ministry of Universities and
2. “Automotive Handbook”, Robert Bosch Gmbh, 1993. Scientific and Technological Research)
3. Hermance D., Sasaki S., "Hybrid electric vehicles take to
the streets", IEEE Spectrum, November 1998.
4. Szumanowski A., Hajduga A., Piórkowski P. "Proper PGS:
Adjustment of Combustion Engine and Induction Motor in
th Planetary Gear Set
Hybrid Vehicles Drive“, Proceedings 15 Electric Vehicle
Symposium, October 1998, Brussels.
5. Wipke K.B., Cuddy M.R., Burch S.D., "ADVISOR 2.1: A PNGV:
User-Friendly Advanced Powertrain Simulation Using a
Combined Backward/Forward Approach", IEEE US Partnership for a New Generation of Vehicles
Transactions on Vehicular Technology: Special Issue on
Hybrid and Electric Vehicles, Columbus, OH, August SIPRE:
1999.
6. Biais F., Langry P. "Optimization of a permanent magnet Sistema Ibrido di Propulsione con Ruotismo
traction motor for electric vehicle", Proceedings 15
th
Epicicloidale (Hybrid-Propulsion System with Planetary
Electric Vehicle Symposium, October 1998, Brussels. Gear Set
7. Bauch-Banetzky D., Rüthlein A., Sonnenburg R.,
"Permanent magneterregte Synchronmaschine mit Nd-Fe-
2WD:
B Dauermagneten für den Einsatz in
Personenkraftfahrzeugen" VDI (Verein Deutscher Two Wheel Drive
Ingenieure) Berichte 1459: Entwicklung Konstruktion
Vertrieb HYBRIDANTRIEBE; VDI Verlag GmbH,
Düsseldorf 1999; ISSN 0083-5560, ISBN 3-18-091459-9. VRLA:
8. Johnson V.H., Wipke K.B., Rausen D.J. “HEV Control Valve-Regulated Lead Acid
Strategy for Real-Time Optimization of Fuel Economy and
Emissions”, Proceedings 2000 Future Car Congress, April
2-6 ’00, Arlington, Virginia.
9. http://leonemar.din.uniroma1.it/ing_avanzamento.html
Use of ADVISOR for simulation of a
Hybrid Electric Vehicle with a
Stirling Engine as the
Auxiliary Power Unit

Owen R. Fauvel and Luis Figueroa

University of Calgary, Canada

Graham T. Reader
University of Windsor, Canada

ADVISOR Simulation of SE HEV

1. Introduction
2. Background
3. Solution Design
4. Current Results
5. Conclusion & Future Work
Questions?

ADVISOR Simulation of SE HEV

• Consideration of environmental impacts in
the life cycle of products
• Environmental regulation & commitment
• Automobile transportation system:
Number of vehicles, driving time
Local by-products: Congestion, air & land pollution
Global by-products: Acid rain, global climate change

• Increasing investment for environmental

improvements on vehicles

ADVISOR Simulation of SE HEV

 (2)
• Hybrid Electric Vehicle (HEV)
– optimization of power units functions:
❏ Acceleration ❏ Cruise ❏ Stand by

• Stirling Engine (SE) powered HEV:

– potential advantages over Internal Combustion
Engines (ICE) powered HEVs

• ADVISOR: Cost effective HEV evaluation tool

– Searching Strengths & Areas of improvement
– Try new ideas

ADVISOR Simulation of SE HEV

• Stirling Engine.
• It operates in a closed regenerative
thermodynamic cycle.
• Cyclic compression and expansion of the
working fluid at two different temperatures.
• Successful use in cryocooling and power
applications.

ADVISOR Simulation of SE HEV

• Stirling Engine.

Southern California Edison (SCE)

Dish Stirling technology ( MDC)
1987, Barstow, California.
13,000 hours, more than 100,000 kWh
http://www.stirlingenergy.com/Pages/technology.html
http://www.kockums.se/productsandserva.html
ADVISOR Simulation of SE HEV
• Stirling Engine: Automotive Applications?

http://www.ott.doe.gov/hev/gmaccomp.html

ADVISOR Simulation of SE HEV

• Stirling Engine: Automotive Applications?

http://www.ott.doe.gov/hev/gmaccomp.html

ADVISOR Simulation of SE HEV

 Hot Space
Regenerator Displacer Piston

Cold Space
Power Piston
P T
3

QRi Qi Qi
4 3 4

2
QRo QRi QRo
2
1
Qo 1

Qo S
V
ADVISOR Simulation of SE HEV
 1 2 3 4
P
T
3
Qi
QRi 4
Qi
2 3 4
QRo QRi
1
QRo
2
1
Qo V
ADVISOR Simulation of SE HEV Qo S
 P 2
3
4
1 T
3
Qi
QRi 4
Qi
2 3 4
QRo QRi
1
QRo
2
1
Qo V
ADVISOR Simulation of SE HEV Qo S
 • ICE • SE
Energy Input
Energy Input 100%
100%

Preheater
Regenerator 43%
430%
Thermal Loss
3%
Exhaust gases
Exhaust 14%
gases
28%
Mechanical
Radiation Loss
Loss 4%
7%
Cooling
Cooling System
System 45% Work Output
33%
Work 34%
Output
32%

ADVISOR Simulation of SE HEV

• Multi fuel capability
• Highest theoretical thermal efficiency
• Higher efficiency than ICEs
• Quieter operation than ICE and gas turbines
• Lower vibration, emission of pollutants than ICE
• Cogeneration capability
• Regenerative braking capability
• Long operating life
• Extended maintenance periods
• Low oil consumption

ADVISOR Simulation of SE HEV

• Reliability
• Higher power to weight ratio (high pressure engines)

• High manufacturing & materials costs

• somewhat larger size than ICE
• Larger radiator than ICE
• Slower response than Otto, similar to Diesel
• Performance = Quality Manufacture
• Catastrophic failures (high pressure engines)

ADVISOR Simulation of SE HEV

• ADVISOR
• MARWEISS
• Emissions calculation

ADVISOR Simulation of SE HEV

 Analyze
Engine ADVISOR
model Stirling
New
Engine File
Engine
Control
Strategy
ADVISOR
Emissions
simulation
Operation Simulation
Parameters

Heat
Transfer
Parameters

SE
Simulation
MARWEISS

Map Generation
(Fuel)

ADVISOR Simulation of SE HEV

ADVISOR Simulation of SE HEV
 GM GPU-3

30

25

20
Efficiency [% ]

15
Efficiency Map
10
46.2919

5 42.0836
37.8752
0 GM GPU-3 33.6669
1 1001 2001 3001 4001 5001 6001 7001 35-40
29.4585
6000 Speed [rpm] 30-35
25.2501
Torque (Nm) 25-30
EFFICIENCY 1.1 [Mpa] EFFICIENCY 2.1 [Mpa] 21.0418
5000 EFFICIENCY 3.1 [Mpa] EFFICIENCY 4.1 [Mpa] 20-25
16.8334
15-20
12.6251
4000 10-15
Power [Watts]

8.4167
5-10
3000 4.2084 0-5
0

100

1975

3850

5725

7600
2000

1000 Speed (rpm)

0
1

01

01

01

01

01

01

01

01

01

01

01

01

01
1
50
10

15

20

25

30

35

40

45

50

55

60

65

70

Speed [rpm]

POWER 0.1 [Mpa] POWER 1.1 [Mpa] POWER 2.1 [Mpa]

POWER 3.1 [Mpa] POWER 4.1 [Mpa]

ADVISOR Simulation of SE HEV

 Preheater Burner Hot Space Regenerator Cold Space Engine Block Hood
A W Cooling Water
E E F F F
C Combustion reactants
Tp C Tb Ths Trg Tcs Te Thd E Exhaust mixture

A Ambient Air

E E W A A F Working Fluid

Ta Tcl Ta
Ambient Coolant Ambient Conduction
W W
Convection
Trd Tht Radiation

Radiator heating

ADVISOR Simulation of SE HEV

 Chemical Energy: Fuel Fuel Converter
80000
70000
60000
50000
40000
30000
20000
10000
0
Fuel Cell

Prius

Parallel
Conventional

Stirling

Stirling
Parallel ICE
Series ICE

Series
ADVISOR Simulation of SE HEV
30.00% Overall Efficiency
Overall Efficiency (PE)
25.00%
20.00%
15.00%
10.00%
5.00%
0.00%

Prius

Stirling

Stirling
Fuel Cell

Parallel
Conventional
Parallel ICE
Series ICE

Series
ADVISOR Simulation of SE HEV
 a

250% 226%
204% 198%
200%
150% 124% 128%
100%
100%
50%

0%
Conventional

Conventional

Conventional

vehicle ZEV
HEV series

HEV series
with ICE

with SE

Electric
with ICE

with ICE

EZEV
with SE
EZEV

a : For a projected production of 10,000 units / year

ADVISOR Simulation of SE HEV

• SE can be an alternative to ICE
– Reduction of emissions
– Improvement of efficiency.
– High engine price: 150% ICE
(with adequate production levels)
• Environmentally friendly automobile:
– Adequate production levels of HEV and SE
– Internalization of environmental costs
• Evaluation of SE as an APU with ADVISOR.
Conventional Series Parallel Split

ADVISOR Simulation of SE HEV

• Experimental testing of the model:
– Efficiency & fuel consumption
– Engine characteristics (Torque, Speed, Power)
– Emissions
– Heat transfer model
• Integrate emissions simulation into ADVISOR
• Evaluate influence of mechanical efficiency on
SE power map generated by MARWEISS.
• Evaluate configurations & components
• Optimize SE HEV.

ADVISOR Simulation of SE HEV

 Questions?
ADVISOR Simulation of SE HEV
 Use of ADVISOR for simulation of a Hybrid Electric Vehicle
with a Stirling Engine as the Auxiliary Power Unit
Luis Figueroa and Owen R. Fauvel
University of Calgary, Canada

Graham T. Reader
University of Windsor, Canada

ABSTRACT global climate change - point to the need to redesign the

current automobile centered-transportation-system. [5]
A hybrid electric vehicle combines the advantages of an
electric motor and an energy storage system, with an Proposed measures in response to the environmental
auxiliary power unit - typically a chemically fueled unit. deterioration include urban planning and widespread use
As a result, it reduces fuel consumption and adverse of public transit and alternative transports such as,
environmental impacts. bicycles and walking. However, it is predictable that the
automobile will remain as the main way of transportation
The aim of this project is to evaluate the performance of in the world major urban areas, urging to find ways to
the Stirling engine use as the auxiliary power unit in reduce its environmental effects.
hybrid electric vehicles and identify the characteristics
that suggest mobility applications of the Stirling engine. In general, the environmental characteristics of vehicles
are not considered as important as other factors, such
A simulation tool for assessing Stirling engine hybrid as security, performance, life style (convenience,
electric vehicles has been designed, based on these equipment), and cost (vehicle purchase price, fuel price)
computer programs: for the average consumer. Government legislation and
(i) ADVISOR, Advance Vehicle Simulator [1, 2] cost reduction are the main driving forces for taking into
(ii) MARWEISS, Stirling Engine simulation program [3] account environmental issues in the design of
(iii) EQUILIBRIUM, Emission composition Simulator [4] transportation. The trend in the automotive industry is to
design vehicles with lower environmental impact and
Results obtained from this designed simulation tool, for increased efficiency, while satisfying customers’
conventional Stirling vehicles or hybrid electric-Stirling expectations.
vehicles in series, parallel or split configurations, can be
compared with other modeled vehicles such as, The automotive industry has developed the technology
conventional, pure electrical vehicles, or hybrid electric needed to achieve the current environmental
powered with internal combustion engines or fuel cells. requirements, paying a price. There is an increasing cost
for the environmental improvements to the customer that
To validate the simulation model, future work will could eventually affect the expected performance of the
compare findings with experimental data from a Phillips vehicle. [6]
Stirling SE 102C power plant and examine findings with
other Stirling engines. The Hybrid Electric Vehicle (HEV) constitutes an
environmentally sound alternative to the conventional
INTRODUCTION automobile due to the optimization of its power units for
acceleration, cruise and standby of the vehicle, leading
Environmental concerns are becoming a determinant to reduced emissions and fuel consumption.
factor in the life-cycle of products (design, production, Furthermore, a Stirling Engine (SE) powered HEV offers
distribution, use and disposal) all over the world. potential advantages over Internal Combustion Engines
Regulation and voluntary commitment to control and (ICE) powered HEVs: the ability to operate on different
mitigate adverse environmental impacts are continuously heat sources (multi-fuel capability), and the reduced
increasing. levels of fuel consumption, noise, vibration and tailpipe
emission related to the Stirling may overtake the
Adverse environmental impacts caused by the intensive projected goals for ICEs.
use of vehicles in big cities are evident. A steady rise in
the number of vehicles and driving time in the world’s A simulation program is a cost-effective tool for
major urban areas is pushing congestion, as well as air performance evaluation of HEVs; it enables the definition
and land pollution to intolerable levels. The by-products of strengths and areas of improvement for achieving
of petroleum combustion - air pollution, acid rain and successful commercialization, reducing costs related to
prototype development.
In this paper is summarized the current status of a Energy Input
graduate research project on the applicability of SE in 100%
HEV applications, to be concluded by December 2000.

The document is organized in the following sections:

Preheater
I. Background. Overview of the SE and HEVs Regenerator 43%
II. Solution Design. Description of the SE HEV 430%
simulation model Thermal Loss
III. Results. Current findings of the project, 3%
comparison of different configurations. Exhaust gases
IV. Future Research. Description of future areas of 14%
research for this project.
V. Conclusion. Recommendations for simulation
and application of the SE HEV. Mechanical Loss
4%

BACKGROUND Cooling System

45%
The purpose of this section is to overview (1) the SE, its
simulation methods and contributions to the automotive Work Output
34%
field, (2) the HEV, the simulator program ADVISOR and
Figure 1. Sankey diagram for energy flow in a typical SE. After
HEV trends.
Walker [11], Reader [12]

THE STIRLING ENGINE

Overview of the SE

The Stirling cycle machine or SE was invented and

patented by Robert Stirling, in Scotland, in 1816 [7]. The Energy Input
100%
SE is a mechanical device, which operates in a closed
regenerative thermodynamic cycle. It is based on the
cyclic compression and expansion of the working fluid at
two different temperatures. The flow is controlled by
volume changes, and there is a net transformation of
heat energy to mechanical work or vice versa [8]. It has
been successfully utilized in cryocooling and power
applications. [9]
Exhaust gases
The Stirling cycle has the same theoretical thermal 28%
efficiency as the Carnot cycle and a larger specific work
output, although this potential has never been achieved.
Radiation Loss
Today SEs present little advantage in terms of efficiency 7%
and power density over ICEs.

The inherent advantages of the SE over the ICE are Cooling System
mainly due to the engine’s continuous external heating 33%
[10]. For comparison purposes the typical energy flow -
diagrams of the SE and the ICE are presented below: Work Output
32%

Figure 2. Sankey diagram for energy flow in a typical ICE. After

Potter[13], Ganesan [14]
Around 70% of the lost energy in a SE is rejected via the 3. Second Order Design Methods; these take most
cooling system, and 21% lost in the exhaust gases, aspects of the SE, and are more adequate for
while in an ICE 50% is rejected via the cooling system design. This methods assume that a relatively
and 40% lost in the exhaust gases. This brings an simple Stirling cycle analysis can be used to
advantage for cogeneration of heat (for the interior of the calculate the basic outputs and inputs, and that the
vehicle) and power with the SE. It also recalls the energy losses can be deducted from the power
requirement for a larger radiator for the SE (usually output; the energy flows are calculated independent
double the size of an ICE radiator). of each other without interaction.

The SE represents a potential alternative to ICE in 4. Third Order Design Methods, or nodal analysis
automobile applications. The main advantages are the: methods simulate the engine by dividing it in a given
§ Ability to use any heating source number of section employing nodes, and solving
§ Highest possible theoretical thermal efficiency numerically the differential equations of
§ Highest possible theoretical mechanical efficiency Conservation of Energy, Mass and Momentum, for
§ Quieter operation than ICE and gas turbines each of the nodes.
§ Lower vibration than ICE
§ Lower emission of pollutants when using fossil fuels, The third order methods are the most accurate and
because of their continuous combustion process appropriate for modeling specific SEs, however, they
§ Cogeneration capability: Heating interior of vehicle require a longer computational and design time than
utilizing the energy kept in the engine’s refrigeration other methods.
system
§ Regenerative braking capability. The engine MARWAISS, a second order method is used in this
operation can be reversed, transforming mechanical project because of its capability to analyze different
power at the output shaft into heat, for deceleration. types of SE within a reasonable computational time. It is
Later on, this heat can be transformed back into validated as a useful computer simulation tool with
shaft power for accelerating the vehicle results close to test bed information, allowing modest
• Higher efficiency than ICEs confidence in the simulation and analysis of energy flow
• Long operating life and efficiency.
• Extended periods between maintenance
• Low lubricating oil consumption MARWEISS or Martini-Weiss is a second order SE
• Reliability (continuous working operation) design aid developed by W. Martini and implemented on
• Higher power to weight ratio when operating at high a PC by M. Weiss at the University of Calgary, Canada.
pressure Based on the Martini isothermal model, performs
decoupled corrections for determining energy losses. It
The main areas of improvement are the: is written in FORTRAN 77, and includes input screens
§ High manufacturing costs with the present low level for the engine parameters. The results of the simulation
of production can be visualized in graphical format and exported to a
§ High materials costs text file. [3]
§ size and weight, somewhat larger than ICE
§ Larger radiator needed compared to ICE, with MARWEISS utilizes the engine general dimensions, heat
closed system cooling exchanger component sizes, drive system dimensions
§ Slower response time compared to Otto engines, and miscellaneous parameters for performing the
similar response as Diesel simulation. It allows the user to redefine a variety of
§ Performance depends on high quality manufacture dimensions, engine configurations, control strategies
§ Failures are catastrophic due to high pressurization and operational parameters.
on engine working fluid
SE Vehicle applications
Simulation of SE
The main thrusts of the modern SE development have
The methods for analysis of SE are classified according been automobile and cryocooling applications. The
to the assumptions and the mathematical theory that history of the modern SE started in the late 1930’s, when
encompass them. They are classified in four main types: Philips researched the development of a 1kW-power
generator. In 1948 Philips designed a 149 kW engine,
1. Zeroth Order Design Methods: simple models based which was seen by Henry Ford II and other major
on an idealized mathematical model (Schmidt cycle automakers in Eindhoven, motivating an interest on the
analysis) and on engine experimental constants engine. Philips used hydrogen and helium for high-
adequate for an initial approximation. speed and high power density SE.

2. First order Design Methods, which start with limited

information for calculating the output and efficiency
of a given SE.
The main projects involving SE for automotive 1981 ranges from $2000/kW to $6700/kW. This cost is
applications are condensed in the following list: practically unacceptable, when compared with the cost
of $25.5 /kW of Otto ICE [12]
§ General Motors-Philips program: From 1958 to
1970. Included GPU-3 6kW (Stir-Lec I series HEV), In order to commercialize successfully the SE in the
4L23 187 kW (bus applications) [15]. Halted due to automotive industry, it must be a standard item, and not
braking problem, high costs and GM team leader a prototype or exception item. Due to the current low
retirement. volume of production, the SE requires specialized
§ Ford - Philips program: From 1970 to 1980. Siemens materials, tools and Fixed Capital Equipment that
- SE double acting 135 kW (Ford Torino). increment drastically the manufacturing cost.
Complemented with Ford - Philips - ERDA program.
§ Ford - Philips - ERDA program: from 1978 to 1980. With a production level of at least 400,000 units per
Siemens - double acting SE 67kW. Halted due to year, according to studies documented by Reader [12],
problems in sizing the engine to the lines of the comparative projected sale price for SE and ICE:
production, and long warming up and cooling down
time of the engine. Table 1. Dealer selling price for automotive engines
§ Stirling Thermal Motors. From 1980 to present.
40kW and STM 4-120 4 cylinder Siemens - Engine 75 kW % Otto 112 kW % Otto
swashplate SE, with hybrid propulsion system, now $/kW $/ kW
team member on General Motors - Stirling Thermal ICE Otto 25.25 100% 18.74 100%
Motors - DOE program. Stirling 39.74 157% 28.92 154%
§ United Stirling AG Sweden: From 1968 to present. 4 1981 USD (400,000 units per year) After Reader, G. Hooper C. (1983)
cylinder 25 to 75 kW (automotive, marine, solar
power generation, and battery charging Nacken Even with less expensive materials and adequate
class submarine of 1200 Tons) Submarine SE with production levels, the SE is likely to be 50% more
an endurance of 6 to 8 times compared to Diesel expensive than Otto Engine. However, it is important to
engine. Automotive application halted due to high consider that because the engine's cost represents
costs. about 10% of the automobile's cost, the replacement of
§ Machinenfabrik Ausburg-Nurnberg (MAN) Motor an ICE with a SE represents an increase of 5% on the
Werke Mannheim (MWM) From 1968 to 1990's. cost of the vehicle.
Double acting Siemens - SEs (Military application on
heavy vehicle engines and underwater power HYBRID ELECTRIC VEHICLES
systems). Halted due to high costs and end of the
cold war. Overview of the HEV
§ Mechanical Technology Incorporated (MTI) -
American Motors Company – DOE. From mid 1970's A HEV is an alternative transport that has two sources of
to late 1980's. 4 cylinder double acting engine with motive energy: 1) an electric motor, and 2) an Auxiliary
crank drive mechanism. Power Unit (APU). HEVs combine both sources with an
§ General Motors - Stirling Thermal Motors - DOE energy storage system (ESS). Typically, the APU is
program. From 1993 to 1998. STM 4-120 4 cylinder chemically fueled -an ICE, gas turbine or a fuel cell.
Siemens -swashplate SE (Partnership for a New
Generation of Vehicles, PNGV). Currently carried As a result of the combination of the power units, HEVs
out. Results have not yet been published.[16] reduce fuel consumption and adverse environmental
impacts. The HEV capitalizes on the principle that a
Several years of intensive research in SE produced vehicle power plant must have three dissimilar functions:
engines that matched the achievements of several i) accelerate, ii) cruise, iii) stand by (when no power
decades of even more intensive research in Otto and output is required: warming up, traffic lights, congestion).
Diesel engines. The position of the SE in vehicle
transportation seems optimistic because of the trends of In a conventional vehicle, the engine is sized to satisfy
the vehicle industry in searching environmentally friendly the acceleration requirement, utilizing excessive fuel for
automobiles, and the natural capability of SE to operate both, cruise and stand by functions. HEVs have two
with low levels of noise, vibration and emission of power plants working on series, parallel, dual or split
pollutants. However, the SE automobile projects have configuration, to:
not been commercialized yet. The main areas for • Optimize power/energy consumption for acceleration
improvement include costs, required time for cooling and • Minimize energy consumption during cruise[17]
warming up of the engine, larger radiator required, • Minimize or eliminate energy consumption during
matching of the engine to the automobile and sealing. stand by
• Minimize pollutant emissions
Cost is the major concern in the SE success; it varies § Provide regenerative braking capability
between 1.5 and 15 times the cost of an equivalent § Allow operation with alternative fuels; need not be
Diesel, according to manufacturers’ data and technical dependent on fossil fuels.
conferences. The cost of a SE prototype in US Dollars of
The advantages of the HEVs over the electrical vehicles unique vehicle components is simple, and a friendly
(EV) are: graphical user interface (GUI) allows manipulation of
§ Lower weight and volume of the electric and energy input files, test routines, and output plots [1,2].
storage units
§ Higher global efficiency of the system, due to direct Cost analysis on HEV
thermal to mechanical energy conversion
§ Better vehicle performance (acceleration and speed) Mercedes Benz conducted a cost analysis for HEV in
due to the motor/APU combination series configuration [21]. The following table contains a
§ Increased autonomy range[18] comparative cost analysis obtained from the study. It has
been complemented with the projected production cost
The disadvantages of the HEVs compared to ICE and for a SE HEV, in series configuration.
EV vehicles are:
§ Higher complexity in design and manufacturing Table 2: Production costs for different drivetrains
§ Higher number of components
§ More sophisticated vehicle electronic control Drivetrain Production
strategies System Cost (%)
§ Higher manufacturing cost Conventional with ICE 100 %
§ Higher weight and volume required for power units Conventional with ICE EZEV 124 %
than ICE. Conventional with SE 128%
HEV series with ICE EZEV 204 %
The HEV powertrain components are selected to HEV series with SE 198 %
minimize its disadvantages and achieve its advantages Electric vehicle ZEV 226 %
without appreciable loss in vehicle performance, range, Based on Mercedes Benz class C (10,000 units per year)
or safety. Complemented after Abthoff, et al. (1998)

Simulation of HEV For calculation of the cost of the SE HEV (PHS):

The methods to analyze HEV employ a static or dynamic • PHIC : HEV ICE price, 204% conventional ICE.
approach for vehicle control to optimize several • PICE : ICE price, 10% price of conventional ICE
parameters, such as efficiency, weight or emissions. • PSE : SE price, for a production level of 10,000 units,
would be around 280% the price of an ICE,
The simultaneous optimization of fuel efficiency and according to studies by Jet Propulsion Laboratory
pollution emissions requires of strong mathematical [22].
algorithms in the computer simulations programs. • PEZEV : EZEV ICE equipment price, given the low
ADVISOR performs forward/ backward analysis to emission of pollutants inherent to the SE, there
review the validity of the information. Other methods would be a reduction of 24% basic vehicle price
currently in study, involve computer models with genetic associated to EZEV ICE equipment.
algorithms that combine billions of options on the
different parameters that determine the engine Then:
performance, for finding better results [19]

The ADVISOR program, launched in 1994 by NREL, PHS = PHIC − PICE + PSE − PEZEV
considers dynamic vehicle operation conditions, for Eqn. 1
optimizing both energy efficiency and emissions, in
conventional and advanced vehicles [20]. It is based on PHS = 204 − 10 + 28 − 24 = 198%
Simulink block diagrams and supported by Matlab data In conclusion, the cost of a HEV with SE would be very
files that contain vehicle configuration, control, and similar to the cost of the HEV ICE that complies with
performance data. The use of Simulink provides EZEV legislation. Environmental concerns themselves
accessible documented code in a graphical environment, are not a reason for the general public to pay double for
which makes modification relatively easy, helping an alternative vehicle.
ADVISOR to be well suited to collaboration between
researchers and for distribution to the public. SE and HEV share a common future: In order to be
commercialized successfully, they must first bring
The code incorporates several drive train types including answers to common problems:
electric, conventional ICE (spark ignited or compression • Need to have adequate production levels to reach
ignited) and HEV in series, parallel and split mode with significant cost reductions in tools, materials and
ICE or fuel cell as the APU. One of the objectives of this fixed capital equipment costs, independently of the
project is to include SE to the drive trains available. APU
• Need to internalize environmental costs in
ADVISOR allows the user to interchange a variety of
transportation pricing [23] [24]
components, vehicle configurations, and control
strategies. Modification of data files to represent new or
The implementation of pollution penalty measures (e.g. SE selection
increased fuel taxes, licensing or driving fees) applied to The engine to be simulated can be selected from
vehicle use and ownership, can motivate the public different sources:
interest towards environmentally friendly efficient a) Engine previously studied with ADVISOR,
vehicles. This will increase the research and production selecting the engine in the graphical interface of
of such vehicles, and a subsequent price reduction. ADVISOR
b) Design of a new engine with MARWEISS. Define
SOLUTION DESIGN a new engine based on i) MARWEISS' engine
parameter definition or ii) edition of an existing
GENERAL SIMULATION MODEL engine
c) Engine previously simulated with MARWEISS,
This section describes the steps considered for selecting the text file with the SE simulation
integrating the simulation tools. results (MARWEISS outputs for the desired
engine)
In order to predict the performance of a variety of SE, d) Input experimental results of the SE in a text file,
the simulation model is flexible allowing to model HEV or for selecting this file later in ADVISOR, as
conventional vehicles, in any configuration. described in option a)

The considerations for the model include: Control Strategy.

• Multi-fuel capability of SE: Simple redefinition of the
type and characteristics of the fuel employed The control strategy is defined by the goal of the
• Control method for SE: Capability for selection of the simulation, the most common options:
control method and generation of the corresponding • Accomplish drive cycle
power maps • Maximum Efficiency
• Reduction of NOx emissions: Consider effects in • Minimum Emissions
power, efficiency and emissions of the recirculation • Minimum Fuel consumption
of combustion exhaust gases for reduced NOx • Maximum Power
emissions. The effects on power, efficiency and • Maximum Torque
emissions should be considered • Maximum Speed

Engine
In ICEs, the power output is a function of the mass
Analyze flows of air and fuel. Because of SE nature, the
Analyze MARWEISS Experimental
Advisor model New Engine
model Data
power output can be controlled in different ways. The
most common ways are:[25]
Control Strategy
1) Pressure level of working fluid
2) Stroke variation
Operation 3) Phase angle variation
Parameters
4) Maximum compression / Expansion volume
5) Temperature of Hot / Cold spaces
Heat Transfer
Parameters The control variable determines the type of power
maps that are required for the simulation. Most
commonly, the power maps will be given as a
Stirling Engine function of:
Simulation
i) pressure level and the angular speed, or
ii) phase angle and the angular speed.
Emissions
Simulation Operation Parameters

Parameters defined to achieve objectives such as,

Map Generation
(Fuel/Emissions) emission levels, power output or efficiency.
1) Energy Parameters
• Type and characteristics of fuel
Stirling Engine Employed
File
• Fuel / Air ratio
2) Engine heater operation
ADVISOR
• Percentage of exhaust gas recirculation
simulation • Ambient temperature and pressure

Figure 3. Methodology for simulation of SE HEV.

Heat Transfer Parameters

These parameters include heat transfer coefficients and

Define Engine
areas of the specific engine, utilized in the simulation of Name
the heat transfer of the engine.

SE Simulation Select System

of Units

As described in the background, it is employed the

second order method MARWAISS. The general
procedure of the program is shown in figure 4. Load
MARWAIS is based on the Martini isothermal method, Yes Engine? No
summarized here:

1. Given the hot and cold space temperatures and the

engine dimensions, calculate basic power using
Schmidt cycle analysis Load Engine File Generate New
Engine File
2. Calculate basic heat input from power output, using
Carnot Efficiency
3. Evaluate net power, net heat input, gas heater duty,
and gas cooler duty
4. Calculate flow rate, cycle time for heater, corrected Select Engine
Configuration
temperature drop for gas heater duty to be
transferred, and effective hot space gas temperature
5. Calculate flow rate, cycle time for cooler, corrected Define Engine
temperature drop for gas cooler duty to be Dimensions and
Parameters:
transferred, and effective cold space gas • Pistons
temperature • Cooler
6. Recalculate steps 1 to 5 using effective hot space • Connecting ducts
• Regenerator
temperature for heat source temperature and • Heater
effective cold space temperature for heat sink • Drive system
temperature • Miscellaneous
• Working Fluid
7. Repeat 6 until the effective calculated temperatures
are steady

Iterative
The power maps generated in MARWEISS tabulate the Calculation
output power and the efficiency as functions of the
engine speed, given different pressures of the working
Display Operating
gas, in the case of engines controlled by working gas Conditions
pressure. For phase angle controlled engines, the power
maps tabulate the output power and the efficiency as
functions of the engine speed, given different phase Power Map
angles. Generation

Stirling Engine
File

Figure 4. Flow diagram for MARWEISS SE simulator

Emissions Simulation 1. Utilizing the values of brake power from the
map, it is determined minimum and maximum,
This part of the procedure consist of the calculation of and generated a brake power vector, with a
the composition and properties of the combustion given number of elements
products at equilibrium temperature, utilizing a C++ code [P]
based on the program EQUILIBRIUM. 2. Utilizing the Matlab interpolation function, from
the brake power map and the pressure vector, it
The calculation is done considering that the products is obtained a mean pressure map as a function
and reactants are at constant pressure, and that the of the angular speed and the brake power vector
reaction takes place at adiabatic flame temperature. It is (generated in step 1)
based in a work by Olikara and Borman [26] further [p=p(w,P)] Eqn. 3
developed by Liensch and Krieger [27], for modeling 3. Utilizing the Matlab meshgrid and interpolating
combustion on ICE at General motors research functions, it is obtained an efficiency map as a
laboratories. Later it was programmed in Pascal function of the angular speed and the brake
language by Lane [4]. For validating the results obtained power vector, utilizing the original efficiency map
from this program, initially the results are being and the mean pressure map, obtained in step 2
compared with experimental information found in the [η=η (w,P)] Eqn. 4
literature for SE, and will be confirmed in the future 4. Given that torque = brake power / angular
experimental work. speed, utilizing meshgrid and the map
generated in (3), it is obtained a brake power
The values of molecular masses, specific heat capacities map as a function of the angular speed and the
and enthalpies of formation, can be taken from a text file torque
containing the Janaf thermochemical tables information [η=η (w,T)] Eqn. 5
for the products and reactants. Alternatively, this
information can be obtained from the NASA polynomial 2) Phase angle controlled engine
interpolation equations.
MARWAISS generates i) brake power map, ii) efficiency
The program EQUILIBRIUM determines the equilibrium map, both as functions of angular speed and phase
composition, as a function of: the fuel / Air ratio, the angle. [P=P(w, α), η=η (w, α)] Eqn. 6
ambient temperature, ambient pressure, type of fuel
utilized and assuming that the reaction will take place in 1. Utilizing the values of brake power from the
ambient air. map, it is determined minimum and maximum,
and generated a the brake power vector, with a
Inputs: Fuel/Air equivalent ratio, ambient temperature, given number of elements
ambient pressure, and number of atoms of Carbon, [P]
Oxygen, Hydrogen and Nitrogen in the fuel molecule 2. Utilizing the Matlab interpolation function, from
the brake power map and the phase angle
Outputs: Average molecular mass, specific enthalpy, vector, it is obtained a phase angle map as a
specific internal energy, moles of fuel, and mole fractions function of the angular speed and the brake
of H,O,N,H2,OH,CO,NO,O2,H2O,CO2,N2,Ar, unburned power vector (generated in step 1)
fuel. [α=α(w,P)] Eqn. 7
3. Utilizing the Matlab meshgrid and interpolating
Map Generation (Fuel/Emissions) functions, it is obtained an efficiency map as a
function of the angular speed and the brake
The power maps generated in MARWEISS, are power vector, utilizing the original efficiency map
transformed utilizing Matlab to the fuel consumption, and the phase angle map, obtained in step 2
power, torque and angular speed maps for simulation on [η=η (w,P)] Eqn. 8
ADVISOR. The methodology for 1) pressure controlled 4. Given that torque = brake power / angular
engine and 2) phase angle controlled engine are speed, utilizing meshgrid and the map
described below. generated in (3), it is obtained a brake power
map as a function of the angular speed and the
1) Pressure controlled engine torque
[η=η (w,T)] Eqn. 9
MARWAISS generates i) brake power map, ii) efficiency
map, both as functions of angular speed and working
gas mean pressure. [P=P(w,p), η=η (w,p)] Eqn. 2
 SE File for ADVISOR
C.V.

Reactants Products This module has as function assembly the maps for fuel
consumption, torque, speed and emissions; writing them
in a text file that ADVISOR can utilize as input file. Also
writes the basic identification of the engine. The output
file of this step can be edited for reflecting any other
required change on the engine.
Burner

Simulate SE HEV in ADVISOR

Q CV W CV
For incorporating the SE in the fuel converter
components of ADVISOR, it is required to define the
Figure 5. Control volume for burner heat transfer model, located under the fuel converter
temperature calculation

HEAT TRANSFER MODEL

The fuel consumption map can be determined from the
second law of thermodynamics, applied to a control Heat is produced by the combustion of fuel in the burner
volume surrounding the burner of the engine: and transferred to the hot space of the engine; it is
transformed into work or transferred to the cold space by
Q CV W CV means of the working fluid. Being SE a thermal engine,
− = hP − hR Eqn. 10 the lower the temperature on the cold space, the better
n F n F for the performance of the engine. However, there will
always be radiation from the different parts of the engine
to the environment, and conduction from the hot space
Q CV to the cold space.
= hPR Eqn. 11
n F
For optimization of the efficiency of the engine, many SE
and designs embody a preheater that works by heat transfer
Q CV from the hot gases of the exhaust of the burner, to the
m = Eqn. 12 air admitted into the burner.
hPR
The calculation procedure starts at the burner,
determining in sequence the heat transferred from a
From Eqn. 12, the fuel consumption map can be given component to the next components, ending with
calculated given 1) the low heating value of the specific the calculation of the heat transfer from the hood to the
fuel, 2) the enthalpies of formation of the products and ambient and from the engine to the cooler.
the reactants, and 3) the efficiency map.
The general equation for calculation of radiation:

1
fc = × 3.6 × 10 6 Eqn. 13 Q x 2 y , r = ε * σ * A * (Tx4 − Th4 ) Eqn. 15
η hPR
For convection:

Q x 2 y ,v = h fluid * A * (Tx − Tamb ) Eqn. 16

Given 1) the molecular fractions of the combustion
products, 2) the moles of fuel and 3) the moles of air
required for combustion, the exhaust gas mass flow can For conduction:
be calculated.

Q x 2 y ,c = hcond * (Tx − Tamb ) Eqn. 17

fc * M i
m i = n
Eqn. 14
åM
k −1
k
 For temperature differences:

Qin. x − Qout . x
t
Tx = ò Eqn. 18
o m *C
x p. x

Preheater Burner Hot Space Regenerator Cold Space Engine Block Hood
W Cooling Water
E E F F F A
C Combustion reactants
C Tb Trg
Tp Ths Tcs Te Thd E Exhaust mixture

A Ambient Air

E E W A A F Working Fluid

Ta Tcl Ta
Ambient W Coolant Ambient Conduction
W
Convection
Trd Tht Radiation

Radiator heating

Figure 6. SE heat transfer model

RESULTS

The simplicity of ADVISOR for simulating vehicles with The ICE employed for the series and parallel
different configurations and components is an advantage configurations, was a Geo 41 kW engine. For the split
over other HEV simulation programs. Once the model of configuration the ICE is a 1.5L Prius (Atkinson cycle)
a component of ADVISOR has been validated, it is engine.
simple and fast to perform simulations.
The fuel cell is an IFC 50kW net hydrogen fuel cell stack.
As an example for evaluation of the modules developed,
it was performed a cycle of simulation routines with The SE employed is a GPU-3 6kW engine. This is a
different APUs in series and parallel configurations, with smaller engine compared to other engines. It was
a basic automobile arrangement. The following table employed in the Sir-Lec I from GM, the first SE HEV [3].
shows the comparison on energy consumption for It was selected for comparison of simulation results with
performing the same test routine, two cycles of the published data.
Federal Urban Driving Schedule.
Some results from the table, are the percentage of
The types of APU and configurations considered are: energy losses on the engine in descending order:
Conventional vehicle (74%), series ICE (69%), parallel
1. HEV ICE in series configuration ICE (69%), split ICE (66%), fuel cell (63%), series SE
2. HEV ICE in parallel configuration (60%) and parallel SE (54%).
3. Conventional ICE vehicle
4. HEV FC vehicle The overall efficiency, including potential energy:
5. HEV ICE in split configuration Conventional vehicle (17%), series ICE (19%), parallel
6. HEV SE in series configuration ICE (20%), fuel cell (22%), series SE (22%), split ICE
7. HEV SE in parallel configuration (23%), and parallel SE (27%).

Table 3 shows the energy balance after each of the The total energy used for performing the cycle: series
components of the vehicle and the percentage of energy ICE (27816 kJ), fuel cell (25666 kJ), series SE
loss in each component, as a fraction of the total energy. (24721 kJ), parallel ICE (22744 kJ), and parallel SE
(22744 kJ), split ICE (21110 kJ), conventional vehicle
(16825 kJ).
The results recall the importance of having a lighter
vehicle to save energy; Even though the conventional
vehicle had the lower efficiency of all the vehicles, it also
used the least energy of all, because it was also the
lighter vehicle.

Table 3: Comparison of drivetrains: Energy use, component

efficiency and overall efficiency.

Energy Usage [kJ] Series ICE Parallel ICE Conventional Fuel Cell Prius Series Stirling Parallel Stirling
Balance [kJ] Use [%] Balance [kJ] Use [%] Balance [kJ] Use [%] Balance [kJ] Use [%] Balance [kJ] Use [%] Balance [kJ] Use [%] Balance [kJ] Use [%]
Fuel Converter 27816 69.2% 22598 69.2% 16825 74.2% 25666 63.3% 21110 66.1% 24721 59.8% 22744 53.8%
Energy Storage 22967 0.2% 22348 0.2% 16252 24375 1.8% 20190 0.9% 22751 3.2% 21701 2.0%
Motor / Controller (Power) 19692 2.5% 20498 2.5% 16252 20957 4.9% 18874 2.1% 18775 5.2% 18889 5.7%
Gearbox (Power) 18662 1.3% 19561 1.3% 15430 1.3% 19892 1.5% 18874 17771 1.6% 17895 2.0%
Wheel / Axle (Power) 17852 1.1% 18747 1.1% 14759 1.0% 19036 1.2% 18067 1.3% 17094 1.1% 17127 1.6%
Motor / Controller (Regen) 17481 1.6% 17549 1.6% 14270 18651 0.6% 17716 0.6% 16733 0.6% 16361 1.6%
Gearbox (Regen) 17412 0.1% 17502 0.1% 14236 0.1% 18582 0.1% 17716 16664 0.1% 16321 0.1%
Aux Loads 15522 2.6% 15612 2.6% 12368 2.9% 16686 2.7% 15827 3.1% 14778 3.1% 14437 3.9%
Braking 14410 1.5% 14492 1.5% 11133 1.9% 15473 1.7% 14545 2.1% 13739 1.7% 13419 2.1%
Aerodynamic loss 12298 2.9% 12380 2.9% 9023 3.2% 13361 3.0% 12894 2.7% 11627 3.4% 11307 4.3%
Rolling Loss 9459 3.9% 9522 3.9% 6940 3.2% 10276 4.4% 9998 4.7% 8966 4.3% 8697 5.3%
Potential Energy 0 13.0% 0 13.0% -1 10.6% -1 0 10 16.1% 8 14.6% 0 17.7%

Overall Efficiency 6.52% 6.78% 6.43% 7.43% 7.31% 7.75% 9.59%

Overall Efficiency (PE) 18.99% 19.78% 17.07% 22.12% 23.37% 22.31% 27.24%

Rolling resistance + aerodynamic losses

Overall Efficiency =
Energy used [fuel + storage system]

Rolling resistance + aerodynamic losses + Potential Energy

Overall Efficiency(PE) =
Energy used [fuel + storage system]

Figure 9. ICE compression ignited power map

Figure 7. ICE spark ignited power map

Figure 8. SE spark ignited power map

Figure 10. ADVISOR results figure.

FUTURE RESEARCH SE automobile projects have not reached
• Experimental testing of the model commercialization in the past. There is a need to match
• Combustion process: Need to review emission the engine to the automobile, look for cost reductions
levels, compared to experimental results and optimize radiator size. Adequate production levels of
• Integration of combustion model into ADVISOR: HEV and SE, and the internalization of environmental
Calculate emissions by time step, given that costs in transportation price, can turn to reality the
emissions change with engine temperature, dream of an environmentally friendly transportation
pressure and Fuel / Air ratio. system.
• Evaluation of the influence of mechanical efficiency
on SE power map generated by MARWEISS. The simulation tool introduced in this document for SE
• Validation of the heat transfer model with HEVs was designed based on ADVISOR HEV simulator,
experimental results MARWAISS SE simulator and EQUILIBRIUM,
• Test of different configurations and components of composition simulator. The purpose of the tool is to
the HEV, for optimizing a SE HEV. evaluate the SE application as an APU, its advantages
and areas of improvement.
CONCLUSION
Results obtained from the designed simulation tool, for
conventional SE vehicles or hybrid electric-Stirling
According to Hitachi Car Eng. Co. (Japan) the ICE will
vehicles in series, parallel or split configurations, can be
remain as the main automobile engine for the following
confronted with other modeled vehicles such as
25 years [28]. One may ask then, if an engine has
conventional, pure electrical vehicles, as well as hybrid
shown such reliability and dominance on the market in
electric powered with ICEs or fuel cells. ADVISOR
such a manner; Why study other options? Considering
allows quick testing of new components and different
the environmental regulation and transportation cost
configurations of vehicles, making the simulation
trends the question would be whether the ICE can meet
process efficient and easy to configure.
all required goals without high increases in purchase
price or complex maintenance requirements.
The next stage of this project is to validate the simulation
model. It is desired to perform experimental studies of
The exhaust gases of automobile engines are polluting
fuel consumption, heat flow, torque and speed
the air in the major urban areas around the world. A
characteristics, as well as emissions for present SE.
steady rise in the number of vehicles and driving time in
the world’s major urban areas is pushing congestion, as
ACKNOWLEDGMENTS
well as air and land pollution to intolerable levels. In
many countries, more than 50% of the air pollution is
attributed to vehicle engines. In some cities, as Los The authors wish to thank the DOE's NREL for
Angeles, it can be even higher. In addition to air pollution providing the opportunity to participate in this
and congestion, the automobile centered transportation international project for searching solutions that
contributes to other sources of annoyance, as noise, generate practical alternative transportation
smell and, in case of diesel, soot. solutions. We extend our special gratitude to the
people that supported and reviewed this document.
The environmental concerns are producing a rethinking
in the world transportation system, in spite of the trend to AUTHORS
increase vehicle cost and the vehicle systems
complexity. Luis Figueroa holds a BS in Mechanical and Electrical
Engineering from Universidad Iberoamericana, Mexico.
From a technical point of view, alternative engines as the He is a graduate student of MS in Mechanical
SE, gas turbine or the fuel cell, can be utilized as a Engineering at the University of Calgary, Canada. He
replacement of the predominant ICE automobile engine, worked in project engineering and in product engineering
for reduction of emissions and improvement of for Dana Corporation Heavy Axle and Brake Division,
efficiency. The main problem that they face, is ICE's and for Eaton Corporation: Truck Components, Axle and
competitive price. SE's multi-fuel capability, reduced Brake Division, from 1994 to 1998. His research looks at
emission of pollutants, and increased efficiency, are simulation of hybrid electric vehicles and the application
overcome by the high price of the engine. Even with of Stirling engines in the automotive field. For further
adequate production levels, it is likely to be 50% more information he may be contacted at:
expensive than ICE, leading to an increment of 5% [email protected]
increment in vehicle price.
Owen R. Fauvel holds a BS in Mechanical Engineering
The HEV is an alternative transport that because of its from the University of Calgary, Canada and a PhD in
double source of motive energy is able to reduce fuel Mechanical Engineering from Newcastle Upon Tyne,
consumption and adverse environmental impacts. The United Kingdom. He has a broad technical background
HEV promotes the optimization of the vehicle in its including experience in the petroleum, aviation,
acceleration, cruise and stand-by functions. construction, and mining industries. His research is
characterized as R&D-focused on the design of Stirling- 12. Reader G.T., Hooper C., "Stirling Engines," E&F.N.
cycle machines, mining/rock excavation. Research on Spon, London, U.K. and New York, U.S.A., 1983.
Stirling-cycle systems has been primarily into the 13. Potter I., "Cogeneration Systems," University of
Calgary, Calgary, Canada, 1998.
modeling and characterization of Ringbom- (hybrid) 14. Ganesan V., "Internal Combustion Engines,"
Stirling engines and of liquid-piston Stirling engines McGraw-Hill, New York, USA,1994.
(Fluidynes). This work has led to the enhancement of the 15. Agarwal P.D., Mooney R.J., Toepel R.R. "Stir-Lec
understanding of their operation, loss mechanisms, as I, A Stirling electric hybrid car," Research
well as giving rise to improved component and system Laboratories General Motors Corporation , U.S.A.,
designs. For further information he may be contacted at January 17, 1969.
16. Anonymous, "Hybrid Electric Vehicle Program, GM
[email protected] Program Overview," US Department of Energy , U.S.A,
Hybrid Electric Vehicle program (HEV), February 26,
Graham T. Reader holds a B.T. and a PhD in 1998, <http://www.hev.doe.gov/>
Mechanical Engineering from the University of Bradford, 17. Taylor, A. III., "Gentlemen, Start your Engines," Fortune,
England, and a BA in Mathematics from the Open p.156-164, September 30, 1996.
University. After 7 years in the mining industry, he joined 18. Caraceni, R. (Fiat Auto-Vama), Caraceni A., Cipolla G.
(Elasis ScPA-Motori), “ Hybrid power unit development for
the Royal Naval Engineering College, United Kingdom.
Fiat Multipla Vehicle, Technology For Electric and Hybrid
He has a broad knowledge in power systems, Stirling Vehicles" 981124, p.29-36, 1998.
and Diesel engines for submarine and environmental 19. Anonymous, "Building Better Engines Through
applications. His expertise on Stirling engines has led Natural Selection," University Of Wisconsin-
him to develop important work, including books and over Madison, U.S.A, June 21 2000.
80 papers. He was head of the department of <http://www.sciencedaily.com/releases/2000/06/000
Mechanical and Manufacturing Engineering at the 621073216.htm> <http://www.wisc.edu/>
20. Johnson V. H., Wipke K. B., Rausen D. J.,"HEV
University of Calgary, Canada. Currently he is the Dean Control Strategy for Real-Time Optimization of Fuel
of Engineering at the University of Windsor, Canada. For Economy and Emissions," National Renewable Energy
further information he may be contacted at Laboratory, U.S.A. 2000.
[email protected] 21. Abthoff J,, Antony P., Kramer M., Seiler J.. “The
Mercedes-Benz C-Class series hybrid,” Technology
For Electric and Hybrid Vehicles, 981123: p.133-
REFERENCES 143, 1998
22. Jet propulsion Laboratory JPL publication 78-71,2.,
1. Anonymous, "Using Systems Modeling to Facilitate the 1978.
PNGV Technology Selection," US Department of Energy , 23. ECTM “Efficient Transport for Europe: Policies for
U.S.A, March 20, 2000, Internalization of External Costs," OECD, Paris, 1998
<http://www.ctts.nrel.gov/analysis/documents/pngv.html> 24. IBI group, Apogee Research Int, Proctor & Refern
2. Wipke, K., Cuddy, M., Burch, S., “ADVISOR 2.1: A User- Ltd., “Inclusion of environmental costs in
Friendly Advanced Powertrain Simulation Using a transportation pricing” IBI group, Canada. Prepared
Combined Backward/Forward Approach,” IEEE for Transport Canada, 1996.
Transactions on Vehicular Technology, Special Issue on 25. Hargreaves C.M., "The Philips Stirling engine,"
Hybrid and Electric vehicles, Columbus, OH, August 1999. Elsevier, Amsterdam, Netherlands, 1991.
3. Walker, G., Weiss, M.H., Fauvel, R. Reader, G. 26. Olikara C., Borman G.L., "A computer program for
"Adventures with MARWEISS: A summary of calculating properties of equilibrium combustion products
experience with Stirling Simulation," University of with some applications to IC engines," SAE paper No
Calgary, Calgary, Alberta, Canada. American 750468, 1975.
Institute of Chemical Engineers, 1990. 25th 27. Liensch, J.H, Krieger, R.B. "A procedure to compute
Intersociety Energy Conversion Engineering composition and gas properties if equilibrium combustion
Conference. (Reno, Nevada, USA, August 12-17, products," General Motors Research Publication GMR-306
1990). March 10, 1981.
4. Lane, N., "A Pascal procedure for the computation of the 28. Ohyama Y., "An Algorithm of optimum torque control for
equilibrium composition and gas properties of combustion hybrid vehicle," 980890 Hitachi car engineering Co, Ltd
products," University of Witwatersrand, South Africa, 1986 SAE Technology for electric and hybrid vehicles, p. 1-
5. OCDE., "Internalizing the social cost of transport" 10,1998
ECMT, Paris, France, 1994.
6. Harmon R., Latam, “Alternative Vehicle-Propulsion
Systems,” Mechanical Engineering: p58-65., N.Y. ADDITIONAL SOURCES
March, 1992
7. Walker, G. Reader, G. Fauvel, O.R. Bingham, E.R. "The 1. Meijer R.J., "Prospects of the Stirling Engine for
Stirling alternative," University of Calgary, Canada, 1990. vehiclular propulsion," N.V. Philips, Eindhoven,
8. Reader G., Potter I., "Introduction to Stirling Netherlands, 1970.
Technology," Canada, University of Calgary, 2. Shaltens R. K. Ernst, William D., "Automotive Stirling
Canada, 1998. engine development," National Aeronautics and
9. Walker, G. Reader, G. Fauvel, O.R. Bingham, E.R. Space Administration; Springfield, Va.: National
"The Stirling alternative." University of Calgary, Technical Information Service, distributor,
Canada, 1990, Washington, DC., 1997.
10. Urieli I., Berchowitz, D.M., "Stirling Cycle Engine 3. Poulton M.L, Southampton, "Alternative engines for
Analysis," Adam Hilger, Great Britain, 1984. road vehicles" Boston: Computational Mechanics
11. Walker G., "Stirling Engines," Claredon Press, Publications, UK, 1994.
Oxford, England, 1980.
4. Anonymous, "Should we have a new engine? An HC:
Automobile Power Systems Evaluation," Jet HydroCarbons
Propulsion Laboratory California Institute of
Technology, USA, 1975.
5. Dochat G., "Stirling engine electric hybrid vehicle HEV:
propulsion system conceptual design study Hybrid Electric Vehicle
[microform]: final report"; Mechanical Technology
Incorporated. Dept. of Energy, Division of HWFET:
Transportation Energy Conservation, Washington, Highway Fuel Economy Test
U.S., 1978.
6. Organ A. J., "Thermodynamics and gas dynamics of
the Stirling cycle machine," Cambridge University ICE:
Press, Cambridge, England, 1992. Internal Combustion Engine
7. Urieli I., David M. Berchowitz B., "Stirling cycle
engine analysis," Adam Hilger, England, 1984. LEV:
8. Anonymous, "Stirling engines : progress towards
Low Emission Vehicle
reality/ conference sponsored by the Power
Industries Division of the Institution of Mechanical
Engineers; with the patronage of Federation NOx:
Internationale des Societes d'Ingenieurs des Nitrous Oxides
Techniques de l'Automobile," Mechanical
Engineering Publications Ltd. for the Institution of NREL:
Mechanical Engineers, London, c1982.
9. Pappas C.H., Murray W.H. III., Osborne, "Visual National Renewable Energy Laboratory
C++ 6: The complete reference," McGraw-Hill,
Berkeley, California, USA. 1998. Phase angle:
10. Sperling, D. Delucchi, M.A., Davis, P. Burke, A.F., Angle position between power piston and compression
"Future Drive Electrical Vehicles and Sustainable piston
Transportation," Island Press, Washington, D.C.
USA.
PM:
Particulate Matter
Definitions, Acronyms, Abbreviations
PNGV:
Partnership for a New Generation of Vehicles
ADVISOR:
Advanced VehIcle Simulator
Preheater:
A heat exchanger user with fuel burners to transfer heat
APU:
from the exiting combustion products to the input
Auxiliary Power Unit
reactant air.
CO:
Regenerative:
Carbon Monoxide
A regenerative engine, is an engine that keeps heat
rejected during part of a cycle in a component named
DOE:
regenerator, for using it in another stage of the cycle
Department Of Energy
Regenerator:
ECE:
Element of a heat engine that keeps heat rejected from a
External Combustion Engine
fluid warmer than itself and releases heat to a fluid
colder than itself
EGR:
Exhaust Gas Recirculation, engine feature for emissions
SE:
reductions
Stirling Engine
ERDA:
SOC:
Energy Research and Development Administration (now
State Of Charge
DOE)
Swashplate:
ESS:
A circular plate centered on and inclined to the axis of
Energy Storage System
plate rotation used to transform reciprocating piston
motion to rotary shaft motion. Usually associated with
EZEV: double-acting engines, but it can be used with single-
Extra zero emission vehicle acting engines.
Fuel / Air equivalence ratio: ZEV:
Zero Emissions Vehicle
GW Hybrid Electric Sport Utility Vehicle Design for
FutureTruck 2000 Competition

GW Transportation Research Institute

Vahid Motevalli, Ph.D., P.E.
Mohamad Syaifuddin
 GW FutureTruck Design Goals

l FUDS and HWFET range - 452 km/300 miles HEV

l ZEV range - 125 km
l Acceleration (0-100 kph) - < 15 seconds
l Emissions - Federal Tier 2
l Equivalent Energy Efficiency - 60 mpg
l Curb Weight - 2,000 kg
l Aero drag, Cd - 0.4
l Gradeability - 5% at 90 kph
l Towing capacity - 900 kg
Estimated Energy and Power Requirement
CRUISING and GRADEABILITY POWER Vehicle curb weight, mvehicle = 2,000 kg
(kW) Aerodynamic drag coefficient, Cd = 0.4
100 kph cruising 19.33 Frontal area, Afrontal = 2.4 m2
Rolling coefficient, Crr = 0.012
100 kph cruising at 6% grade 54.25
Rotational inertia coefficient, Cinertia = 1.2
65 kph cruising 7.93
65 kph cruising at 6% grade 30.63
100 kph with 900 kg load 22.28

100 kph with 900 kg load at 6% grade 71.92

ACCELERATION POWER
(kW)
Peak power for 0-100 kph in 15 seconds 116.96
ENERGY
RANGE STORAGE
(kWH)
Energy: 120 km ZEV range at 65 kph 14.64
 Vehicle Component Layout

Figure 1. GW FutureTruck vehicle component layout

Powertrain:
Charge Sustaining Thermostat Series
Motors: 1 90 kW DC electric motors at rear axle
Engine: VW 1.9 liter TDI
Motor/Generator: 1 90 kW motor/generator
Energy Storage System (ESS): NiMH batteries and ultracapacitors
VW TDI specifications:
-Four-cylinder in-line diesel with direct
injection turbocharged (TDI)
-60 kW at 3,300 rpm
-Engine management system with EDC
-Sealed pressurized cooling system with
oil cooler in coolant circuit
-Height: 637 mm, Width: 710 mm,
Length: 545 mm
-Dry weight = 135 kg
 Model MDF-375/75-A1-BS
NGM's 90 kW motor characteristic MDF-375/75-A1-BS specifications:
-Design Topology: Axial Flux, Dual Stator Fixed Air gap
800
-Overall size: 445 mm diameter x 189 mm length
700
-Weight: 84 kg
600
-Cooling: Forced Air
Torque (Nm)

500
-Mounting: Face flange per outline
400
300
-Max rated current : 320 amps
200
-Rated DC Bus voltage: 312 V
100 -Mechanical speed limit: 3,000 rpm
0 -Rotor inertia: 0.17 kg-m2
0 500 1000 1500 2000 2500 3000
Motor speed (rpm)
 Ni-MH & Ultracapacitors

NiMH battery stack specification Ultracapacitor stack specification

Battery pack Ovonic Ultracapacitor Pinnacle
specification NiMH specification RIT Inc.
[ 30 kWh] [ 130 kW]
No. of modules 25 No. of modules 3

Mass (kg) 445 Mass (kg) 1.3

Volume (liters) 188.3

Volume (liters) 1

Power@20% SOC 97.8

(kW) Specific power 100
(kW/kg)
Voltage (V) 330
Voltage (V) 300
 Power Requirement During Acceleration

Power curves vs time • High power demand > 120 kW

(Acceleration: 0-60 m ph) • NiMH batteries provides only
140 maximum of 90 kW
120
• Ultracapacitor provides
Pow er_NiMH required power for acceleration
100
Power (kW)

Pow er_Ultracap
80
Pow er_required
60
40
20
0
0 5 10 15 20
time (s)
 Simulation Results
HWFET Cycle

HWFET cycle (kph) ESS_SOC vs time (1 cycle HWFET)

120 1.02
100 1
0.98
80

ESS_SOC
0.96
Speed (kph)

60 0.94
40 0.92
0.9
20
0.88
0 0.86
-20 0 200 400 600 800 1000 0 200 400 600 800 1000
time (s) time(s)

Power curves vs time (1 HWFET)

60

50
Po
wer 40
(kW
) 30 Prqd
20 Pregen

10

0
0 200 400 600 800 1000
-10
time (s)
 Simulation Results
FUDS Cycle

FUDS cycle (kph) ESS_SOC vs time (1 FUDS cycle)

100 1.02
90
1
80
70 0.98
Speed (kph)

ESS_SOC
60 0.96
50
40 0.94
30 0.92
20
0.9
10
0 0.88
0 500 1000 1500 0 500 1000 1500
time (s) time (s)

Power curves vs time

70

60 Pregen
Prqd
50

40
Power (kW)

30

20

10

0
0 200 400 600 800 1000 1200 1400 1600
-10
Time (s)
 Summary

l The combined fuel economy prediction is 39 mpg (35.5 gasoline

equivalent).
l The use of motor-generator at the front axle is unique and offers
benefits of cost saving and mass reduction.
l The use of two transaxle motors allow four wheel drive mode without
any need of transmission and transfer case.
l The hybrid energy storage system of NiMH batteries and
ultracapacitors allows vehicle to achieve high performance and
increase its range.
l Further study is required for component integration and modeling of
the hybrid energy storage system.
 GWU Hybrid Electric Sport Utility Vehicle Design for
FutureTruck 2000 Competition

Mohd-Syaifuddin Mohd, Zeki Gokce, Vahid Motevalli, Kartik Bulusu

GW Transportation Institute, The George Washington University

out the necessity of addressing all aspects of vehicle

ABSTRACT
effeciency improvements as itemized here.
The George Washington University is one of
the several universities invited to participate in the
Table 1 - GWU FutureTruck design goals
FutureTruck 2000 competition. The competition
challenges engineering students to convert a Sport PARAMETER TARGETS
Utility Vehicle (SUV) to meet the goals of attaining up
to three times their current fuel efficiency, achieving FUDS and FHDS range 462 km/300 miles HEV
exhaust emissions of Federal Tier 2 levels or below, and ZEV range 125 km
maintaining consumer acceptability in areas of
performance, utility, and safety. In preparation for the Acceleration (0-100 kph) < 15 seconds
competition, a design of a Hybrid Electric SUV was
Emission Federal Tier 2
developed and evaluated. The evaluation of conversion
of SUV was conducted by the GWU FutureTruck Equivalent Energy 60 mpg
Team, focusing on the powertrain configuration, heat
Efficiency
engine selection, energy storage strategy, control
system strategy and electric drivetrain components. Curb weight 2,000 kg
This paper describes the process, an innovative series
hybrid powertrain configuration, and discussion of Aero drag, Cd 0.40
simulation performance characteristic of such a vehicle. Gradeability 5% at 90 kph
Even though GM Suburban has been designated as the
platform vehicle, this analysis was performed for a Towing capacity 900 kg
general compact SUV.

INTRODUCTION To improve fuel efficiency, the team has

decided to propose an innovative series hybrid
To achieve the FutureTruck 2000 goals, the powertrain configuration. Details about the
GWU Future Truck Team has set forth several goals configurations are discussed in the following sections.
which are listed in Table 1. The magnitude of these With the selected configuration, potential mass
tasks can only be met through significant revision of the reduction can be made by removing the existing engine
entire SUV system. Five technical targets are identified block, transmission, and part of the driveshaft. The
which are: team has also considered replacing the existing Heating,
Ventilation, and Air-Conditioning (HVAC) system with
• Improve the fuel efficiency of the primary fuel Thermoelectric Cooling (TEC) HVAC system. The
converter TEC system is more efficient, capable of cooling and
• Reduce the mass of the vehicle heating, using less power, has no CFC dependence and
is lightweight compared to to the existing compression
• Reduce acessory system power loads based system [4]. Other potential mass reduction is also
• Improve vehicle aerodynamic efficiency and possible through the use of composite material to
reduce other dynamic losses replace the original body panels and hood. The
aerodynamic performance of the vehicle can be
• Implement regenerative braking improved by several measures: changing the vehicle
profile and eliminating points where separation occurs,
Previous work on conversion of conventional vehicles
reducing discontinuity along the body line by inserting
into Hybrid Electric Vehicle (HEV) by other
rubber trimming and covering the rear wheels, and
universities [e.g. 1,2,3] under the Department of
eliminating the irregular underside surface by installing
Energy’s (DOE) Future Car Challenge program, the
a floor pan. Table 2 shows the estimated energy and
Partnership for a New Generation Vehicles (PNGV)
power requirement for the HEV-SUV based on the
program and the HEV program in DOE have all pointed
targets described before. The total power demand OVERALL DESIGN AND SYSTEM
equation represents the energy consumption of a vehicle INTEGRATION
over a period of time. The equations used to calculate
The team selected a charge sustaining series
the power demand are [5]:
hybrid powertrain configuration. The benefits of a
Ptotal = Paero + Proll + Phill + Pacceleration (1) series configuration over a parallel configuration are
3 [6];
Paero = ½ ρairAfrontalV Cd (2)
i. the engine never idles, which reduces vehicle
Proll = mvehiclegVCrr (3) emissions and increases fuel efficiency,
Phill = mvehiclegVsinθ (4) ii. the engine drives a generator at an optimum
Pacceleration = arequiredmvehicleVCinertia (5) performance setting, and
where; iii. a series hybrid does not need a transmission,
which will reduce the vehicle mass.
Ptotal = total required power
Two highly efficient 90 kW DC motors will
Paero = power needed to overcome aerodynamic loss drive the proposed vehicle, one at the front axle and the
Proll = power required to overcome rolling resistance other at the rear axle. The motor will be used to capture
energy during regenerative braking. The front motor is
Phill = power required to climb hills used alternately as a driving motor when coupled to the
Pacceleration = power required to achieve desired frontal drive shaft or as a generator when coupled to a
acceleration Volkswagen 1.9 liter compression-ignition direct
injection (CIDI) engine. Figure 1 shows the vehicle
Table 2 - Estimated energy and power requirement component layout. An electromagnetic (EM) clutch
system is used to engage and disengage the front motor
CRUISING AND GRADEABILITY POWER to the drive shaft or the IC engine. Power to the motor
(kW) will be provided by a hybrid battery pack consisting of
100 kph cruising 19.33 Ovonic Nickel Metal-Hydride (NiMH) batteries and
Pinnacle Research Institute, Inc. ultracapacitors. A
100 kph cruising at 6% grade 54.25 motor controller from New Generation Motors
65 kph cruising 7.93 Corporation will be used to control the desired speed or
desired torque of the motor. An onboard CPU unit will
65 kph cruising at 6% grade 30.63 be used as a Powertrain Control Module (PCM) which
100 kph with 900 kg load 22.28 stores the control strategy algorithm of the vehicle.

100 kph with 900 kg load at 6% grade 71.92

HEAT ENGINE SELECTION
ACCELERATION POWER
(kW) A 1.9 liter Volkswagen Compression Ignition
Direct Injection (CIDI) engine has been selected for
Peak power for 0-100 kph in 15 seconds 116.96
this design. It is a four-cylinder in-line diesel engine
with turbocharger and direct injection. The CIDI
ENERGY engines were considered for their high thermal
RANGE STORAGE efficiency (>40 percent), operating flexibility, low start-
(kWH) up emissions, and demonstrated manufactureability and
Energy: 120 km ZEV range at 65 kph 14.64 affordability, reliability, durability [7]. CIDI engine are
also capable of increasing fuel economy by up to 35
percent [7]. Furthermore, in the series configuration that
Vehicle curb weight, mvehicle = 2,000 kg has been designed here, some of the undesirable effects
Aerodynamic drag coefficient, Cd = 0.4 known for the older designs in CIDI and not completely
resolved in the current engines, such as particulate and
Frontal area, Afrontal = 2.4 m2
CO emission as well as noise and vibration is
Rolling coefficient, Crr = 0.012 minimized since the engine would be operated in a
Rotational inertia coefficient, Cinertia = 1.2 nearly constant rpm at an optimum setting (highest
efficiency) and with least amount of transient problems.
The selected VW engine has a good specific power with
a compact design. Figure 2 shows the engine
characteristics.
 Figure 1. GW FutureTruck vehicle component layout

ultracapacitors are that it takes less than 1 second to

fully recharge the unit and also minimizes the battery
size. The challenge is in coupling the ultracapacitors to
the battery pack. Table 3 and 4 show the specification
for the NiMH battery stack and the ultracapacitor stack,
respectively.

Power curves vs time

(Acceleration: 0-60 m ph)
140
120
Pow er_NiMH
100
Power (kW)

Pow er_Ultracap
80
Pow er_required
60
40
20
0
Figure 2 - Characteristics of a 1.9 liter VW CIDI 0 5 10 15 20
engine [8] time (s)

ENERGY STORAGE DEVICE SELECTION Figure 3. Power curves vs time during hard
acceleration
The GW team has decided to use a hybrid
configuration of Ovonic NiMH batteries and Pinnacle Table 3 - NiMH battery stack specification
Research Institute, Inc. ultracapacitors to achieve the Battery pack specification Ovonic NiMH
power and energy storage requirements of the vehicle. [ 30 kWh]
During a short period of high power demand,
ultracapacitors will provide the necessary power to the No. of modules 25
motors. We project that the NiMH battery stack alone
will not be able to provide the high power required for Mass (kg) 445
hard acceleration (see Table 2) and would exceed the Volume (liters) 188.3
targeted mass limit. This has dictated the need of an
ultracapacitor pack which has 90 times the power Power@20% SOC (kW) 97.8
density of the batteries [9]. Other advantages of using
Voltage (V) 330
Table 4 - Ultracapacitor stack specification
ELECTRIC DRIVETRAIN
Ultracapacitor specification Pinnacle RIT Inc.
[ 130 kW] The drivetrain will consist of two prototypes
90 kW New Generation Motors Corporation (NGM)
No. of modules 3 DC motors. Each one of the motors will drive the front
drive shaft and the rear drive shaft. During low to
Mass (kg) 1.3 average power demand, only the rear motor is driving
Volume (liters) 1 the vehicle and the front motor is in the generator mode.

Specific power (kW/kg) 100 When the driver demands hard acceleration,
the EM clutch system will disengage the generator from
Voltage (V) 300 the engine and engage it to the frontal drive shaft. The
power circuit is now reversed and has turned the
generator into electric motor mode and will drive the
frontal drive shaft.
CONTROL SYSTEM DESIGN AND ITS
COMPONENTS
NGM's 90 kW motor characteristic
The control system will consiste of power
controllers for the electric motors and the on-board
800
computer system also called a Powertrain Control
Module (PCM). The new NGM-EV-C200 controllers 700
will be used to control and power the 3-phase DC 600
motors. The controllers have programmable logic
Torque (Nm)
500
capability allowing them to be optimized and matched 400
with the motor characteristics. The overall system
300
controller will be implemented on the on-board PCM.
200
This computer will be designed to monitor the state of
charge of the batteries and ultracapacitors, individual 100
motor voltage and current, vehicle speed, temperature 0
of various components, generator output, and will have 0 500 1000 1500 2000 2500 3000
various control functions including activation of the Motor speed (rpm)
CIDI engine and generator, EM clutch engagement or
disengagement, cooling fans, power limited cruise
control, and motor controller. Figure 4 - NGM’s 90 kW DC motors characteristic

The control strategy utilized is the thermostat The prototype NGM 90 kW DC motor is
series, which has the objective of operating mainly on chosen because of its high efficiency (>90%), high
electrical energy. The battery will be discharged until a specific power (each weigh only 55 kg) and capability
low state of charge (SOC) is reached. The engine is to operate efficiently in regenerative mode. Figure 3
then turned on to recharge the battery until a desired shows characteristic of the motor.
SOC is achieved. Advantages of this type of control
strategy are that the engine can be set to operate at one
point of torque and speed at which it is most efficient FUEL SYSTEM DESIGN AND MATERIALS
and least polluting. This prevents the engine from SELECTION
experiencing any transient load, which also should The proposed vehicle will utilize the existing
reduce the emissions. In addition, the operating point fuel system with only minor modifications to
can be set for the optimum performance point of the accommodate the new Volkswagen 1.9 liter CIDI
electric generator by using a simple gear coupling. engine. This has several advantages in that the fuel tank
Another potential reduction to emissions can come from is optimally placed for crashworthiness, the fuel lines
the addition of an electrically heated catalyst, which can and connections are commercially produced and meet
be easily implemented to reduce cold-start emissions all federal standards, and the integration into vehicle is
since there is ample knowledge of when the engine will already achieved. This minimizes the need to make
come on [10]. modifications in the layout and focuses the integration
effort on new components. The only difference in the
fuel system is that the flow rate is no longer regulated
by the gas pedal and fuel pump, but rather will maintain
a constant flow rate to the generator engine and will be
activated by a signal from the PCM. The existing fuel
tank can also be replaced with a smaller tank since the
conversion vehicle will require less fuel to achieve the Assuming a 10% decrease in aerodynamic
same driving range as the original vehicle. drag, an improved rolling resistance of 0.012,
Exhaust Emissions Control Design and Component maintaning vehicle mass of 2,000 kg (model year 1999
Selection Chrysler’s Jeep Cherokee [13]) and using a series
hybrid configuration, the results show that the vehicle is
The catalytic converter, muffler, and the capable of achieving more than 39 mpg fuel economy
tailpipe will have to be resized to accommodate the new (35.5 gasoline equivalent). This fuel economy
CIDI engine. The design will address emission control projection is more than two times that of a typical SUV.
to meet the Federal Tier 2 as required by the With the addition of efficient regenerative braking and
competition rules. Since the engine will operate at a control system design algorithm, and mass reduction,
constant speed for charging purposes only, high this design vehicle has a potential of reaching the 3
emissions typical of today’s vehicle due to acceleration, times current fuel economy. Table 5 shows an example
deceleration, and time at idle are virtually non-existent. of simulation parameters and conditions used to predict
vehicle performance and energy economy projection.
PERFORMANCE AND ENERGY ECONOMY Table 5- Simulation parameters
PROJECTION Simulation Set-up Parameters
Variables
The GWU team has used ADVISOR , a
Drive cycle HWFET (Highway Fuel
MATLAB-based dynamic simulation software to Economy Test Drive)
predict performance and energy economy projections of No. of cycles 25 (256 miles)
the proposed vehicle [11,12]. However, the software Initial SOC of battery pack 95 percent
could not fully simulate the proposed design of the
Maximum SOC 95 percent
hybrid energy storage system of NiMH batteries and
Minimum SOC 30 percent
ultracapacitors. The simulation model does not include Control Strategy Zero delta SOC Correct
a ultracapacitor stack as specified in the vehicle design. (SOCfinal = SOCinitial)
This would not affect the energy economy prediction
since the maximum power demand during the
simulation of FUDS and HWFET cycles is below 50
The use of two 90 kW motors and an
kW and it is achievable by the NiMH battery pack alone
ultracapacitors stack allows the vehicle to meet the high
(see Figures 5 and 6).
power demand during hard acceleration.
Power Requirement (FUDS cycle) Ultracapacitors although having high power density,
70
havw a limited energy storage capacity. Therefore, a
hybrid configuration of NiMH batteries and
60
ultracapacitors is proposed to meet the 300 mile range
50 requirement. The greatest challenges will be in design
40 and implementation of the EM clutch to use the front
Power (kW)

30
motor as the motor/generator and the coupling of the
ultracapacitors and battery stack.
20

10

0 SUMMARY
0 200 400 600 800 1000 1200 1400 1600
-10
Time (s) The GWU innovative series hybrid conversion
approach has a potential to meet and exceed the design
Figure 5- Power requirement vs time for 1 FUDS cycle goals outlined for FutureTruck 2000 competition. The
Power Requirement (HWFET cycle) use of highly efficient and compact DC motors as
generator and driving motors is unique and offers
60 benefits of cost saving and mass reduction. The use of
50 two transaxle motors also allows the vehicle to operate
40 in four wheel drive mode. The hybrid energy storage
system of NiMH batteries and ultracapacitors makes it
Power (kW)

30
possible for the vehicle to achieve high performance
20 and increase its range. A highly efficient CIDI engine
10 will provide power to the generator only in its high
0 efficiency region and low emissions. In addition, non-
0 200 400 600 800 structural body parts will be replaced with lighter
-10
composite materials and the vehicle aerodynamic drag
time (s)
will be reduced through several measures.
Figure 6- Power requirement vs time for 1 HWFET
cycle
CONTACT
12. Senger, R.D., Merkle, M.A. and Nelson, D.J.,
‘Validation of ADVISOR as a Simulation Tool for
Vahid Motevalli, Ph.D., P.E. Series Hybrid Electric Vehicle’. Society of
GW Transportation Research Institute, Automotive Engineers, 1998
The George Washington University,
13. DaimlerChrysler Corporation, ‘Jeep Grand
20101 Academic Way, Ashburn, VA 20147
Cherokee Service Manual’, 1998.

REFERENCES

1. Higdon, D. and Marcinkoski, J., ‘1998 University

of Maryland FutureCar Design Report’, Society of
Automotive Engineering, Annual Congress and
Exposition, Detroit, MI, 1998.
2. Maguire, P., Heimburg, J.T., Nickchen, D.J. and
Sullivan, A.M., ‘Design of Charge Regulating,
Parallel Hybrid Electric FutureCar’, Society of
Automotive Engineering, Annual Congress and
Exposition, Detroit, MI, 1998.
3. Alexander, M., Johnston, B., Carlson, C., Moran,
B., Walter, G., Carlson, R., Kucera, P., McClory,
M., Yee, R., Meyr, N., Frank, A. and Burke, A., ‘A
Mid-Sized Sedan Designed for High Fuel Economy
and Low Emissions: The UC Davis FutureCar’,
Society of Automotive Engineering, Annual
Congress and Exposition, Detroit, MI, 1998.
4. Tellurex Corporation, ‘Advanced Thermoelectric
Cooling Devices’. Available at the Corp. web site,
http://www.telurex.com/resource
5. Bosch, R., ‘Automotive Handbook’, 4th edition
Society of Automotive Engineers, 1996.
6. U.S. Department of Energy, ‘Hybrid Electric
Vehicle Program’. Avalaible at
http://www.hev.doe.gov.
7. U.S. Council for Automotive Research,
‘Partnership for New Generation of Vehicles:
Technical Accomplishments in Detail’. Available at
http://www.uscar.org/pngv/technical/index.htm.
8. Volkswagen of North America, Inc., Industrial
Engine Division, October 1996.
9. Pinnacle Research Institute, Inc., ‘Ultracapacitors:
Origin, Status, and Future’. January 1999
10. U.S. Department of Energy,‘Request for Proposals
for FutureCar 2000’, January 1999.

11. Burch, S., Cuddy, M., Johnson, V., Markel, T.,

Rausen, D., Sprik, S., Wipke, K., ‘Advanced
Vehicle Simulation: Advisor 2.1 Documentation’,
Avialable at the National Renewable Laboratory’s
web site,
http://www.ctts.nrel.gov/analysis/advisor_doc/,
updated April 1999.
 Simulation of a Heavy-Duty Series
Hybrid Electric Transit Bus
Presentation
Paper
Jean Bavard and Atef Gayed, Alstom Transport

Hybrid Buses - The Benefits of Matching

to Real Routes
Presentation
Paper
Mike Kellaway and Alan Ponsford, Newbus Technology Limited

ADVISOR Modeling of a Fuel Cell

Hybrid Electric Vehicle
Mike Ogburn and Doug Nelson, Virginia Tech
 August 2000

Simulation of a Heavy-Duty
Series Hybrid Electric Transit Bus
Jean BAVARD / Atef GAYED
 INTRODUCTION

l ALSTOM & Urban Transport

l HEV Motivations
l Design Criterion
l Software Simulation via ADVISOR
l Results
l Conclusion

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 ALSTOM & URBAN TRANSPORT

- TROLLEYBUS ATHENS - - TRAMWAY BORDEAUX -

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 Electric powered bus
600 / 750 V DC line

Electrical propulsion drive diagram

Internal 350V DC line
Line converter
Braking Braking
resistor chopper 350V part HV part

Motor Motor
Control 80 kW 80 kW
Alternator Rectifier
supervisor

Excitation CAN link Inverter Inverter

Energy Inverter Inverter

Storage

Bus control Motor Motor

80 kW 80 kW

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 ALSTOM & URBAN TRANSPORT

GOALS
l HYBRID ELECTRIC BUS
– Factory tests 2002
– Prototype validation 2003
– Mass production 2004

l FUEL CELL BUS

– Demonstrator 2002

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 HYBRID ELECTRIC VEHICLES
MOTIVATIONS
l GOALS :

– Pollutant Emission Reduction

– Fuel Consumption Reduction
– Better Comfort

l Means :
– Energy Storage System
– Energy Management
– ZEV operating range
– Electric Traction

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 BUS DESCRIPTION
l 12 meters long
l 18 tons max. weight (15.2t for simulations)
l Performances :
– easy starting up to 18% grade,
– around 20 km/h for 15% grade,
– 50 km/h with 5% grade,
– 82 km/h on flat road
– acceleration max : 1.5m/s2
– Jerk max=1.4 m/s3 >>>> (Comfort)

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 DESIGN CRITERION

l Series Hybrid Electric Vehicle

– Maximize Energy Recuperation Possibility
– ZEV Possibility.
– Modular Assembly Possibility (1 or 2 axles)
– Easy Fuel Cell integration
l Strategy
– Maximize the electrical motor use in the Braking Phase
– Optimize Engine Operating points

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 ENERGY STORAGE SYSTEM
l Novel Battery Technology allow low SOC
operating point
– cs_hi_soc=.6 / cs_lo_soc=.4
l Sized to accept :
– “Max.” Regenerative Power
l Zero Delta SOC
l ZEV
– Battery Pack with High Power will have Relatively
Large Energy Content.

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 Electrical Motor & ADVISOR

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 Electrical Motor & ADVISOR

Original ADVISOR traction/brake

Modified brake effort and power
effort and power

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 Electrical Motor & ADVISOR
l MC Bloc Diagram

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 WHEEL & AXLE
l Maximize electrical braking
l Mechanical Brakes :
– Very Low speed range
– Braking Torque needed exceeds Electrical motor
Torque

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 CYCLES
Speed (Km/h)
FUDS
CYCLES FUDS TCL21E
_AR
Max decel (m/s2 ) 1.475 1.4
Max. Spd (km/h) 91.2 45
Ave. Spd (km/h) 30.9 14
TCL21E_AR Time (s) 1372 7332
Distance (km) 11.8 28.7
# of stop/km 1.6 4.1

Time (Sec.)

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 RESULTS : FUDS / ST
Energy Usage Table (kJ)
POWER MODE REGEN MODE
In Out Loss Eff. In Out Loss Eff.
ADVISOR

Energy Storage 46113 34430 11616 0.75

Energy Stored 67
Motor/Controller 53951 45939 8013 0.85 22940 21978 962 0.96
Gearbox 45939 44203 1735 0.96 23069 22940 130 0.99
Final Drive 44203 44203 0 1 23069 23069 0 1
Wheel/Axle 44203 42321 1882 0.96 25618 25311 307 0.99
Braking 2242

ADVISOR_MODIFIED
Energy Storage 47157 35184 12027 0.75
Energy Stored -54
Motor/Controller 53934 45939 7995 0.85 24817 23008 1809 0.93
Gearbox 45939 44203 1735 0.96 24951 24817 134 0.99
Final Drive 44203 44203 0 1 24951 24951 0 1
Wheel/Axle 44203 42321 1882 0.96 25618 25279 339 0.99
Braking 328

Gain Regen_Motor_in = 8.2% Gain Mechanical brake=85.37%

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 RESULTS : FUDS / Regen Mode
CYCLES FUDS (S T) FUDS (S FO)
ma x_ve h_bra ke _pwr (kW) 266.5 266.4
Ma x_we h_bra ke _pwr (kW) 220 221
ve h_re ge n_ e ne rgy (kJ ) 25618 25618
Motor_re ge n_e ne rgy_in (kJ ) 24817 24817
Motor_in/ve h_re ge n_in e ff. 96.87% 96.87%
Motor_re ge n_e ne rgy_out (kJ ) 23008 23399
Motor_out/ve h_re ge n e ff. 89.8% 91.3%
Friction bra ke e ne rgy (kJ ) 328 328
Ga in in Friction bra ke los s e ff 85.37% 85.37%
Fue l cons umption (l/100km) 35 36.2

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 MC_160/230kW Operating Points

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 CONCLUSION
l Thanks to ADVISOR Teamwork
– User facilities (modifications, adaptation, …)
– Friendly User Interface
– but difficulties to know how a variable is used
in order to make modification

l Continuing with ADVISOR :

– Optimize pollution reduction
– Optimize Control Strategy
– Add interfaces with ALSTOM design tools

- Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus -

 Simulation of a Heavy-Duty Series Hybrid Electric Transit Bus
Jean BAVARD
Hybrid Electrical Vehicle Research & Development Manager
ALSTOM Transport -Tarbes- France
Phone : +33.(0)5.62.53.48.69
E-mail : jean [email protected]

Atef GAYED
Research & Development Engineer

ABSTRACT 1. The reduction of the mass, tire rolling, aerodynamic

and transmission losses receives great effort from
Using ADVISOR simulation tool, the aim of this paper is the automobile engineers.
to present a study concerning a series hybrid electric
vehicles (S-HEVs) with both thermostatic and load 2. Optimization of each individual component efficiency
follower control strategies. The vehicle being studied is a of the system. This objective may be realized by
12 meters, 18 tons transit bus. First, the weight the properly selecting the component type, size,
regenerative braking energy will be outlined. The configuration and control strategy. Thus, all
importance of the electrical motor and the energy components work together in order to achieve the
storage system and their impact on recovering this best overall efficiency required.
energy is highlighted secondly. Finally, this paper
presents a comparison between the ADVISOR original 3. Recovering the kinetic and potential energies that
version and a modified one that takes into account the are usually wasted in the form of heat in the brakes.
real speed/torque map in the braking operation mode. This is an important approach to improve the vehicle
Improvements due to the new braking version are efficiency especially in the city part of the bus duty
presented. cycle where stop and go driving is frequent.
INTRODUCTION
Both point 2 and 3 are central goals of the EVs and
In urban areas, vehicles accelerate and decelerate HEVs research in general and ALSTOM in particular.
frequently. As conventional engine drive vehicles waste
all of their inertia energy as heat in braking, energy EV & HEV REGENERATIVE BRAKING
efficiencies from fuel to vehicle traction are typically less
than 15% [1]. Therefore Electric Vehicles (EVs) and We can even say that both points have to be optimized
Hybrid Electric Vehicles (HEVs) which have re- together in order to achieve better efficiency and
generative braking feature are suitable for urban use autonomy of EVs and HEVs. In fact, when braking,
from energy saving point of view in addition to the fact
that they are zero or low emission possibilities. The 1. Only the driven axle can be regenerative. Another
amount of the re-generative energy that is fed to traction part of the braking energy will be dissipated as heat
battery is one of the key parameters to evaluate the EVs by the friction braking system of the non-drive axle.
and HEVs. Batteries or supercapacitors from one hand
and the electric motors [2] on the other hand have an 2. If the braking power exceeds the capacity of the
important role in recovering this regenerative braking regenerative braking system, the amount of the
energy. recovered energy is influenced by the part of the
braking energy absorbed by the frictional brake
Using ADVISOR, the aim of this paper is to present a system.
study concerning a series HEV transit bus with both
thermostatic and load follower control strategies. The 3. The recovered and recycled energy will be
effect of the regenerative energy will be outlined. influenced by the losses due to the inherent physical
properties and limits of each component used to
EV & HEV EFFICIENCY produce and store this energy.

4. The braking energy recovery in emergency situation

Improving the energy efficiency of a vehicle can be that can reach a deceleration up to 0.6g is not
made by three ways: significant. This paper will focus on normal urban
driving cycle.
SYSTEM DESCRIPTION

PERFORMANCES AND ELECTRICAL MOTOR

The series hybrid bus under study is 12 meters long and

weighs 18 tons. Figure 1 shows the Alstom AC induction
motor traction characteristics used during this study. It
presents the sum of two hub motors of 80kW each
implemented in a dual-mode bus (trolley or
diesel/generator fed electric drive). It shows also the
torque needed at different speeds in order to achieve
performance for flat and different road grade (5, 10, 15
and 20%). The motor speed/torque characteristics Figure 2: Electrical motor braking characteristics
achieve easy starting up to 18% grade, around 20km/h
ENERGY STORAGE SYSTEM (ESS)
for 15% grade and 50km/h with 5%grade. Naturally it
achieve 82km/h on flat road.
The energy storage system (type, size, configuration and
its control strategy) is one of the most important
components in EVs and S-HEVs in order to improve fuel
economy, overall efficiency and emission reduction.

We had used a novel battery technology during this

study. It was sized in order to respect two criteria:

First, to be able to receive "all" the driveline braking

power.
Second, to accept the genset power, when it is on, in the
same time with the driveline braking power. (i.e. that the
ESS can ensure alone the traction performances).

Respecting these criteria will lead to:

Figure 1: Electrical motor traction characteristics
1. Oversizing the battery pack. Generally, with the
Figure 2 shows the amount of the vehicle braking power chemical batteries available today having the
with respect to its speed for different deceleration. required high power will also have a large energy
Alstom has chosen to realize & use an AC induction content.
2
motor that decelerates up to 1.5m/s and can recover up
to 230kW. The deceleration criteria had been taken 2. Have a zero emission possibility on the bus duty
symmetrical with the maximum passenger comfort cycle (ex: down town) depending on the energy
acceleration. Alstom can meet, with other motor, greater content of the battery pack.
deceleration in function of the bus specification. It must
be noted that the motor torque/speed curve in the 3. Freedom to choose the genset operating point,
braking operation mode is different than that the traction 4. Better regenerative motor efficiency.
side as assumed by ADVISOR. Consequently, this will
lead to another maximum regenerative power than that 5. Improved fuel economy
of the traction power (160kW) as shown in figure 1 & 2.
GENSET
Figure 2 shows also that the maximum braking power is
proportional to the vehicle speed up to a key point "A ". The genset consists of the internal combustion engine
Beyond this key point the regen_motor braking power (ICE) or Fuel Converter (FC) together with a generator
stays constant for a while and then drops. and power electronics (Generator Controller GC).

In general, by pushing the key point "A" to the left Genset is the set of power/speed combinations for a
(minimizing the speed@maxpower), we can have better given vehicle where the maximum efficiency lies, where
bus performances and thus better energy recuperation. power and speed are limited by either the engine or the
generator.
ALSTOM second stage program will be with a
Permanent Magnet Synchronous Motor (PMSM) which The FC is a diesel engine, which have 93 kW maximum
will improve efficiency and regenerative power at high power and 4000 rpm maximum speed, with 38%
speed range. efficiency at 290Nm@2200rpm

The GC is a speed generator with a maximum torque of

290Nm, speed range up to 4000rpm.
WHEEL AND AXLE DRIVE CYCLES (CYC)

The wheel drive line brake fraction (Wh_dl_brake_frac) Figure 4 presents the FUDS, TCL21E_AR urban drive
and the wheel friction brake fraction cycles. The last one represents a French drive cycle in
(Wh_fric_brake_frac) are shown in Figure 3. These Lyon City. These cycles have high dynamic variation.
assume that the maximum brake force will be applied to The maximum speeds are approximately 3 times the
the vehicle by the electrical motor in order to recover the average speed as shown in Table 2.
maximum admissible of the braking power. The
mechanical brakes will do the rest. Speed (Km/h)

FUDS

TCL21E_AR

Figure 3: Distribution braking force on the front axle

AUXILIARIES: Assuming a continuous 6kW electrical

TRANSMISSION: 1 fixed speed

Time (Sec.)
VEHICLE: 12m, 18t. Simulation results take into account Figure 4: FUDS & TCL21E_AR urban driving cycles
that consumption is usually calculated for half loaded
bus of 15.2tones.
CYCLES FUDS TCL21E_AR
MODELING & SIMULATION
2
max decel (m/s ) 1.475 1.4
Using ADVISOR, the component information presented
above are the input data files for the S-HEV bus max_speed (km/h) 91.2 45
structure. This part must be designed as well as
possible knowing the performances to be met. average_speed (km/h) 30.9 14

CONTROL STRATEGIES PARAMETERS (CS) time (s) 1372 7332

Distance (km) 11.8 28.7

Control strategies parameters are presented in the Table
1 with zero delta SOC target met for all cases studied: Numbers of stop/km 1.6 4.1
cs_parameters S. Thermostat S. Follower.
Table 2: FUDS & TCL21E_AR cycle data
cs_min_pwr 46kW 10kW
CASE 1 : FUDS WITH ADVISOR ORIGINAL FILES
cs_max_pwr 46kW 70kW
The ADVISOR original version assumes that the
cs_lo_soc 0.4 0.4 electrical motor has the same maximum torque curve in
traction and braking operation mode (dotted lines Figure
cs_hi_soc 0.6 0.6 5,6). Running the simulation with the above input files
and conditions gives us the following results shown in
Soc_init 0.5 0.5
Table 3 and Figure 5. We can note that the electrical
motor is used over all of its speed and torque map
Table 1: cs_parameters range. The mc_trq is limited at high speed in traction and
The new battery pack technology used has at medium and high speed in braking while it is not
approximately constant Open Circuit Voltage (OCV) over limited at low speed. The vehicle has a maximum
a wide SOC range. This allows a low SOC target. braking power of about 266 kW. The wheel power is
Consequently the energy storage system cs_parameters limited at the motor maximum admissible power
are chosen in order to recover the maximum admissible (151kW).
from the regenerative motor energy.
 Figure 6: ADVISOR & real motor brake power

Figure 5: Motor controller operation for FUDS cycle with

original ADVISOR version

The total vehicle braking energy is 25618 kJ. The

motor_in kJ 22940 for both control strategies. The
motor_in/veh_regen energy is about 89.5%. The motor
can send to the battery pack 21978 kJ (series
thermostat - ST) or 22027 kJ (series follower - SFO).
The regen efficiency is about 86% and the brake losses
are 2242 kJ.

CYCLES FUDS (ST) FUDS (SFO)

max_veh_brake_pwr (kW) 266.5 266.4

Figure 7: Motor controller operation for FUDS cycle with
max_weh_brake_pwr (kW) 151.1 151.1 modified ADVISOR version
veh_ regen_energy(kJ) 25618 25618
CYCLES FUDS (ST) FUDS (SFO)
Motor_regen_energy_in (kJ) 22940 22940
max_veh_brake_pwr (kW) 266.5 266.4
motor_in/veh_regen_in eff. 89.5% 89.5%
max_weh_brake_pwr (kW) 220 221
Motor_regen_energy_out (kJ) 21978 22027
veh_regen_ energy (kJ) 25618 25618
motor_out/veh_regen eff. 85.8% 86%

Friction brake loss energy (kJ) 2242 2242 Motor_regen_energy_in (kJ) 24817 24817

Fuel consumption (l/100km) 35.5 36.8 motor_in/veh_regen_in eff. 96.87% 96.87%

Table 3: Regen simulation results for FUDS cycle for both motor regen_energy_out (kJ) 23008 23399
series control strategies with ADVISOR version
motor_out/veh_regen eff. 89.8% 91.3%
CASE 2 : FUDS WITH ADVISOR MODIFIED FILES
Friction brake energy (kJ) 328 328
The ALSTOM AC induction motor used in this study has
another speed/torque map, Figure 6, in the regenerative Friction brake loss eff . 85.37% 85.37%
operation mode. This allows it to absorbe 70kW
(dot line) more than in traction mode especially in the Fuel consumption (l/100km) 35 36.2
medium and high speed range. Introducing this
difference enabled us to approach the real motor
operation and refine the simulation study and lead to the Table 4: Regen simulation results for FUDS cycle for both
following results. series control strategies with ADVISOR modified version
Table 4 and Figure 7show that while braking : strategy with the electrical motor and the energy storage
system characteristics allow the recuperation of a very
L the mc_trq is not limited
large amount of the braking power.
L the wheel driveline regen power can accept up to
221kW First, the importance of the braking energy for EVs
L brake losses decreased by 85.37% for both control & HEVs was highlighted. Optimizing each individual
strategy. component of the system from one hand and recovering
the kinetic and potential energies on the other hand are
L the regen to motor has increased by 8.18%
the central goals of research for these structures. In the
L the motor energy sent to the battery pack has same time, the electrical motor and the energy storage
increased by 4.6% (ST) and 6.2% (SFO). system characteristics are of major importance to
L Slight improvement in fuel economy. The improve the regenerative recuperation. Consequently,
cs_parameters have to optimise in order to have a introducing the real motor speed/torque map in the
real gain for such point. regenerative operation mode have allowed us to
minimize the mechanical brake losses and thus improve
CASE 3: TCL21E_AR the regenerative energy efficiency as presented for the
urban FUDS and TCL21E_AR cycles.
Table 5 presents the simulation results for a thermostatic
control strategy in the same conditions and parameters ACKNOWLEDGMENTS
for the TCL21E_AR Lyon urban cycle. The first column
presents the original ADVISOR version while the second Special thanks to the ADVISOR team works specially
is with the modified one. We can see the same Valerie JOHNSON for their precious help.
conclusion as found for the FUDS cycles. The difference
is the amount of energy recovered. The gain in the CONTACT
mechanical brake is about 52% instead of 85.3%for the
FUDS cycle. This can be explained by the fact that the
Jean BAVARD was born in Dijon, France in 1958. He
TCL21E_AR cycle has half maximum and average
received the B.Sc. degree in 1980 in general and
speed than the FUDS.
electrical engineering from the Ecole Centrale de Paris,
CYCLES TCL21E_AR TCL21E_AR France. Since 1982 he joined the electrical traction
department of Alstom-transport in Lyon and participated
ADVISOR Adv_Modified in the development of variable speed drives with
alternative machines applied to railways and marines.
max decel (m/ss) 1.415 1.4
Currently, he is in charge of traction drives applied to
max_veh_brake_pwr (kW) 236 236 hybrid electrical bus. The main fields of interest are
system control, energy management and power
max_weh_brake_pwr (kW) 151 222.75 electronics.

veh energy (kJ) 41574 41574 Atef GAYED was born in Ismaïlia, Egypt in 1964. He
received the B.Sc. degree in 1986 in power electrical
motor regen energy_in (kJ) 38989 40018 engineering from Zagazig University, Cairo, Egypt. He
received the Ph. D. degree in 1996 from Nantes
motor_in/veh_regen eff. % 93.78% 96.25% University, France in the application of sliding mode
control to variable speed drives. Since 1999, he
motor energy_out (kJ) 35763 36003
participates as engineer for Alstom-transport in the
motor_out /veh_regen eff. % 86% 86.6% railway and hybrid electrical bus fields. His research
interest includes electrical motor control system, energy
Friction brake loss energy (kJ) 2018 967 management and power electronics.

Fuel consumption (l/100km) 35.4 35.2 REFERENCES

Table 5: Regen simulation results for TCL21E_AR cycle 1. Yimin Gao, Liping Chen and Mehrdad Ehsani,
'Investigation of the Effectiveness of Regenerative Braking
for series thermostat control strategies for both original & for EV and HEV', SAE 1999-01-2910
modified ADVISOR version
2. Mellor, P.H. 'High Efficiency Drivbe-trains for Electric and
CONCLUSION Hybrid Vehicles' IEE 1999.

In urban driving cycles, the friction brakes use a

significant amount of energy. A well matched braking
APPENDIX

Figure 1: Electrical motor traction characteristics...............................................................................................................................2

Figure 2: Electrical motor braking characteristics...............................................................................................................................2
Figure 3: Distribution braking force on the front axle.........................................................................................................................3
Figure 4: FUDS & TCL21E_AR urban driving cycles .......................................................................................................................3
Figure 5: Motor controller operation for FUDS cycle with original ADVISOR version ...................................................................4
Figure 6: ADVISOR & real motor brake power .................................................................................................................................4
Figure 7: Motor controller operation for FUDS cycle with modified ADVISOR version ..................................................................4

Table 1: cs_parameters........................................................................................................................................................................3
Table 2: FUDS & TCL21E_AR cycle data .........................................................................................................................................3
Table 3: Regen simulation results for FUDS cycle for both series control strategies with ADVISOR version ..................................4
Table 4: Regen simulation results for FUDS cycle for both series control strategies with ADVISOR modified version...................4
Table 5: Regen simulation results for TCL21E_AR cycle for series thermostat control strategies for both original & modified
ADVISOR version ................................................................................................................................................................5

ACRONYM & VARIABLE LIST

T (Nm) Torque

P (kW) Power

Adv_M_br_T (Nm) Advisor Motor braking Torque versus motor speed

M_br_T (Nm) real Motor braking Torque versus motor speed

M_br_P (kW) real Motor braking Power versus motor speed

SOC State Of Charge

ST Series Thermostat

SFO Series Follower

TCL21E_AR Transport en Commun Lyonnais (ligne 21 Aller Retour)

Hybrid buses - the benefits of matching to real routes

Mike Kellaway Alan Ponsford

1 ADVISOR Conference 2000

Background - UK market

Dart

Classic Dart SLF

2 ADVISOR Conference 2000

 Future Bus Requirements
Environment

Vehicle

Passenger Operator

3 ADVISOR Conference 2000

Broad Scope

On-board
systems

Vehicle

Infrastructure

4 ADVISOR Conference 2000

 Challenges

ØEnergy use Saving 16 to 60%

ØEmissions reductions
ØTrue systems engineering
ØInfrastructure links
ØReal benefits to whole society

5 ADVISOR Conference 2000

 Series Hybrid Bus

ØEnergy source flexibility

ØZEV range possibility
ØCompatible with realistic battery capability
ØOffers both CO2 and pollutant improvements
ØPackaging flexibility
ØFacilitates links to infrastructure
ØMatch to urban application

6 ADVISOR Conference 2000

 Performance envelope required

ØModern city profiles

ØAttractors of low speed (energy, noise, pedestrians)
ØBlock time (gate to gate) is important parameter
ØOpportunities to match vehicle to route

7 ADVISOR Conference 2000

 Real routes ?

ØObjectives of standard cycles

ØMeasured data
ØReal variability

8 ADVISOR Conference 2000

 Real routes
R o u t e m e a s u re d in L o n d o n
35

30

25

20
m ph

15

10

-5
0 500 1000 1500 2000 2500 3000 3500
S e c s s in c e s t a rt

9 ADVISOR Conference 2000

 Data gathering

ØLimited data available from operators

ØGPS
ØReal-time GIS review during initial trials
Ø3-axis accelerometers/grade estimation
ØCustom logger with error-based filtering

10 ADVISOR Conference 2000

 Scatter

Payload
Traffic
Driver
Seasonal
…

mph

11 ADVISOR Conference 2000

 Modelling Strategy

ØCreate model of existing conventional bus

ØExercise on collected route data to benchmark approach
ØDevelop hybrid bus model as a primary design tool
ØDevelop performance envelope reduction strategy

12 ADVISOR Conference 2000

 Why ADVISOR?

ØOpen source format

ØLink to Matlab used for data processing
ØAvailable models
ØExcellent support resource available

13 ADVISOR Conference 2000

 Modelling issues

ØFluid flywheel
ØPayload
ØIntermittent accessory load
ØRWD
ØOptimisation

18 ADVISOR Conference 2000

Reduced-performance envelope possibilities

19 ADVISOR Conference 2000

 Findings
ØADVISOR simulation gave good fidelity compared with
real-world fuel consumption for conventional design

ØReduced envelope designs offer significant and

achievable benefits

ØADVISOR promotes insight into the key issues in

optimising a design

20 ADVISOR Conference 2000

 Hybrid Buses – the benefits of matching to real routes
Mike Kellaway and Alan Ponsford
Newbus Technology Limited

ABSTRACT cost of all of the drivetrain components including the

battery. There is also a significant efficiency benefit in
Existing city buses are able to cover a wide operational avoiding the use of a large heat engine at low power
envelope in terms of speed, acceleration and grade. levels. Reducing vehicle unladen weight further reduces
This significantly exceeds the requirements of many the power requirement. In a future commercial
routes particularly in flat, busy cities. The purchase and environment, where fuel use is likely to become more
operating costs of a conventional bus are not particularly heavily penalised, reduced fuel consumption will clearly
sensitive to performance level, but this is not true for a become important.
hybrid bus. Limiting the performance envelope of the bus
can significantly reduce the size, weight and cost of the As with any optimised vehicle, there is clearly a danger
major systems on the vehicle. However it is essential to of unexpected performance shortfall. It is therefore
fully understand the real route requirements and to use necessary to collect representative data about actual in-
accurate simulations to optimise the vehicle service route dynamics and its variability. This then
specification. This is being done using on-route forms a qualified basis for careful simulation during
measurements and ADVISOR simulations. vehicle design. It is only through working with statistical
data during the simulation process that one is able to
INTRODUCTION gain insights into the 'best' specification and to
demonstrate the practicality and benefits of reduced
Newbus Technology Limited are developing innovative envelope designs.
hybrid buses based on a thorough evaluation of the
vehicles' requirements and whole system design.
Vehicle dynamic performance is one area being closely
looked at.

Throughout the world, city buses are generally specified

to be capable of relatively high speeds and climbing
severe grades. In the UK, a typical modern vehicle can
maintain around 50 mph on the level and climb a 25%
grade fully laden. Clearly these are excessive
capabilities in many city applications where traffic
speeds are generally low and steep grades are not
encountered. A typical London bus for example, such as
that shown in Figure 1, will not exceed 30 mph in
operation on almost all routes. The level of power
available on the more recent designs is indeed sufficient
to achieve longitudinal acceleration levels that can cause
passenger discomfort, especially for standing
passengers, if the vehicles are not driven carefully.

Because the purchase cost of a conventional bus is not

heavily influenced by its power level, and in many
markets fuel cost is not a prime concern, there has been
little impetus to reduce power levels. In fact, drivers
have expressed a strong preference for highly powered
vehicles. There is also some concern that buses can
accelerate quickly to match traffic speed.

In the design of a hybrid vehicle there is increased

Figure 1 Picture of London bus on route
motivation to reduce the peak power level of the vehicle,
as this has a significant influence on the size, weight and
Measured data also provides a solid basis for 'what-if' DATA COLLECTION METHOD
investigations of more radical proposals such as limiting
driver authority. Initial studies have been based on data collected for
London Bus by Millbrook. These have allowed overall
POTENTIAL FOR A REDUCED ENVELOPE BUS route dynamics and scatter on one typical route to be
identified. These data were obtained from an
Because a bus has a relatively fixed duty, there is scope instrumented vehicle in service.
to produce a range of vehicles with the performance of
each model tailored to a class of routes. Whilst there To make it practical to collect what is likely to become
are undoubtedly operational considerations such as a large amounts of data it was important to develop a
preference to be able to switch any bus to any route, in robust and low-impact data collection approach. It is
reality there are many cases where a vehicle remains on also attractive to achieve the required level of accuracy
one type of route throughout its life. However, without excessive cost, particularly as it may be
commercial considerations make it attractive to necessary to monitor a significant number of vehicles in
investigate whether the distribution of route requirements parallel. Finally, it was considered important to be able
would support a single initial reduced envelope design to collect the data remotely both to save time and to
that would cover a reasonable number of route minimise the effect of such data collection on driving
applications whilst showing significant benefits. This can style.
only be done by measurement of route characteristics.
The approach taken is to use a vehicle-mounted GPS
It must be emphasised that there is no attempt to design receiver to collect 3 dimensional fix and time data. This
a bus specific to a single route, rather a 'route class- is supplemented by 3-axis accelerometers that can be
specific' vehicle. Having said this, the Newbus designs used to provide some data during any loss of GPS
will incorporate adaptive features to allow automatic signal, for example in 'urban canyons' or under bridges.
tailoring to individual routes within the class. The accelerometers used are able to read gravitational
force and can therefore be used to measure vehicle
IN-SERVICE DATA REQUIREMENTS inclination when the vehicle is stationary or moving with
constant velocity. The data is processed to give a 'most
Although it is clear that many routes offer scope for a likely' set of readings before being stored in flash
reduced performance envelope design, it is necessary to memory. A radio modem will be used to allow remote
collect a significant sample of on-route dynamic data to data access.
allow the potential benefits to be studied by simulation.
This can be divided into route topological characteristics GPS data are recorded once per second, the
and statistical on-route dynamics. accelerometer data are measured more frequently and
filtered down to be recorded at the same rate as GPS
ROUTE TOPOLOGY data. This has the benefit of allowing peak accelerations
to be captured and improves accuracy when integrating
The gradients found on a route, their length and the to estimate velocity and displacement.
speeds at which these are driven have a strong
influence on both peak power and intermediate energy DATA COLLECTION EXPERIENCE
storage requirements for hybrid designs. It is therefore
important to characterise a selection of routes in this Initial trials of the equipment have been encouraging,
respect. Initially, measurements are being made to particularly the horizontal data. This was checked
establish typical cases and some idea of spread across against a GIS map on a portable PC in real-time during
the routes in a city. This allows the likely benefits to be initial data collection to allow the quality of the data to be
identified before more extensive characterisation is assessed and to investigate the type of features that
carried out. caused difficulty. In general a reasonable number of
satellites were in view, even in city streets and the only
OPERATIONAL STATISTICS difficulties came when passing under bridges. The route
included a section along some relatively narrow London
It is also essential to investigate the distribution of streets and there was some evidence of loss of precision
dynamics on a single route caused by varying traffic caused by reflections from tall buildings. This is not
conditions and as a result of different driving styles. It is thought to be a problem as such minor events can be
important to understand the causes and scale of such processed out of the data. Although the results
real-world scatter. This is being carried out by confirmed that achievable vertical GPS repeatability is
measuring vehicle dynamics on selected routes for an quite good for measurements taken close in time, this is
extended period and identifying the variations within this. not sufficiently good to estimate grade. The use of full
The objectives are to establish what are acceptable differential GPS is being considered, but may not offer
performance levels, how prevalent is 'over-driving' and enough improvement to be worthwhile.
how significant would be the effects on route timings of
performance limitations.
 The accelerometer pack gave good results which are Following initial trials it was decided for future work to
clearly independent of the GPS data quality, however it carry out an initial 'topology' measurement on each
proved difficult to reliably isolate gradient data from route, where precautions could be taken to ensure good
longitudinal accelerations on normal vehicle runs. grade information was obtained by stopping frequently
Further work is under way to improve these along the route. The statistical data can be collected
measurements. Figure 2 illustrates the speed variation quite effectively in-service as planned. The
along the initial route measured. Figure 3 illustrates the GPS/accelerometer pack has proved itself effective for
variation in speed distribution from different runs on the both types of measurement.
same day.
CHOICE OF ADVISOR
Route data used for simulation
35
ADVISOR was chosen as the simulation tool for this
30 project for a number of reasons:

25 Ø Its open-source format allowed visibility of the

simulation process and ease of adaptation to
20
[Figure 3 Typical data traces] specific requirements
Speed (mph)

15
Ø Processing of measurement data was performed in
10
Matlab and was therefore easily linked to ADVISOR.

5 Ø The wide range of models available to allow work to

be quickly started
0

-5
0 500 1000 1500 2000 2500 3000 3500
Secs since start

Figure 2 Measured route data used for simulation

Figure 3 Fraction of time in different speed (mph) bands for

consecutive runs
ADVISOR SIMULATION STRATEGY HYBRID VEHICLE

CONVENTIONAL VEHICLE A hybrid simulation model was then created based on

the Newbus hybrid design and used to investigate the
The initial route data was input as a cycle to an performance of alternative hybrid designs.
ADVISOR model with component models reflecting the
design of a typical London diesel bus as shown in Figure ROUTE DATA
1. The fuel consumption derived from the simulation
(5.52 mpg) compares well with actual values that are in The use of real route data is preferred to standard cycles
the 5-6 mpg range for this type of route. The as the design is being optimised for representative
simulation's level of achievement of the demanded routes rather than to a particular standard. This is more
(measured) speed profile was also compared with likely relevant for fixed-route vehicles such as buses. It is also
load patterns during the day and confirmed that the important to include the effects of variability between
simulation was giving a realistic model of the vehicle. drivers and different traffic conditions. An example of
Figure 4 illustrates the match at the selected payload. this is the change in speed profiles as the bus moves
The ADVISOR results give fuel consumption in miles per into the city centre.
US gallon rather than the figures above, which are in
miles per UK gallon.

Figure 4 Results for conventional vehicle

ADVISOR MODELS PAYLOAD

Initial vehicle models are derived from the built-in The payload was set by running the model with
ADVISOR sub-system models, suitably modified to increasing load until the best match with the measured
reflect the conventional and Newbus vehicle designs. speed was achieved. In reality the payload changes
As the development progresses it is anticipated that with time and at some time it would be interesting to look
more of the elements will use Newbus-specific models. at changes in this as part of the simulation as it has a
These will be particularly important for the Newbus major influence on fuel consumption and the
custom hybrid controller, which incorporates a number of performance required.
new features, and the battery model, which is to include
a battery damage model to assess in-service battery ACCESSORY LOAD
lifetime. Some particular modelling issues arose during
the work: It would be helpful to be able to model varying accessory
loads as these can vary with engine speed and/or
FLUID FLYWHEEL strategy. Without this the load tends to be overstated on
the average or understated at the peaks. It is planned to
The conventional bus uses a fluid flywheel rather than a capture real operational data in this area.
torque converter and this was modelled as a manual
gearbox with slightly higher numerical ratios in the lower
gears combined with reduced overall efficiency.

Figure 5 Reduced envelope hybrid results

REAR WHEEL DRIVE CONCLUSIONS

It would be helpful to put this in explicitly. Even though Initial results confirm the potential for a lower power
this is almost certainly immaterial here, it would be hybrid design that offers lower purchase and operating
comforting to know that everything was the correct way costs, and reduced emissions in comparison with higher
round. power hybrid designs. Ongoing work will develop this
approach and allow the optimum power level to be
FUEL CONVERTER 'SWEET SPOT' specified taking environmental and commercial
considerations into account.
Some effort was taken to match the generator and fuel
converter to bring the fuel converter operating point to The use of in-service dynamic measurements coupled
maximum efficiency at low operating speed. This may with ADVISOR simulation is essential to allow this
be possible in a more efficient way with more familiarity approach to be carried out with confidence.
with ADVISOR.
By actually capturing measurements from vehicles in
MOTOR/CONTROLLER SPEED OPTIMISATION service it was possible to validate the data on a
conventional vehicle before moving to a hybrid design.
The optimisation tool was found to be very useful in
setting overall reduction ratio. The Newbus reduced envelope hybrid shows a reduction
in fuel consumption of almost 40% over a conventional
INITIAL FINDINGS bus. This is achieved with battery cycling designed for
long service life.
Initial results from the first ADVISOR hybrid models are
shown in Figure 5. Significant fuel savings were
possible in comparison with a 'full-envelope' hybrid
design. As is generally recognised, it is necessary to ACKNOWLEDGMENTS
carefully match the individual elements of the hybrid
driveline to deliver the potential of each hybrid design. The authors would like to acknowledge the assistance
given with access to existing route data by Mike Weston
The initial simulations confirm that a reduced envelope of London Buses and Andy Eastlake of Millbrook Proving
vehicle could service all sections of the route measured Ground. We would also like to thank the ADVISOR
in London. In many cases the acceleration levels team for their advice and response to requests for
currently used offered no savings in route timing, but additions to ADVISOR, in particular the addition of
clearly reduced passenger comfort and increased fuel gradient input to the model.
usage and emissions. There is therefore scope for
further reductions by further reducing performance, CONTACT
though this clearly depends on the results of ongoing
route measurements. Mike Kellaway can be contacted at [email protected],
Alan Ponsford at [email protected]. The Newbus
NEXT STEPS Technology Limited website at www.newbus.com also
contains further material about our work on Hybrid
The data collection technique is to be used in a number Buses.
of UK cities to collect a representative sample of route
data. This will then be characterised and used as
simulation data for a new reduced power hybrid design.
The use of carefully chosen actual route data will allow DEFINITIONS, ACRONYMS, ABBREVIATIONS
detailed assessment of the operation of the new vehicle
and optimisation of its control strategy. GPS:

The process of in-service measurement coupled with Global Positioning System

detailed ADVISOR simulation will allow a thoroughly
researched requirements specification to be drawn up
for the vehicle with a high level of confidence of GIS:
successful operational performance.
Geographical Information Software
Future simulation work will include emission data and
cover the spread between routes, drivers and time of
day in more detail.
Hybrid Electric Vehicle Team of Virginia Tech

Mike Ogburn
Mechanical Engineering
Virginia Tech
Hybrid Electric Vehicle Team of Virginia Tech

2nd Place Overall and Lowest Emissions, 1999 FutureCar Challenge

Hybrid Electric Vehicle Team of Virginia Tech

• Vehicle Layout and Design

• Fuel Cell Subsystems
• ADVISOR Model Development
• Experimental Data vs. ADVISOR Results
• Modeling of Proposed System Changes
• Conclusion
Hybrid Electric Vehicle Team of Virginia Tech

•Power density (kW/kg)

similar to IC engine

•Efficiency roughly twice

that of an IC engine

•Zero emissions
capability

•Low operational Energy Partners NG2000

maintenance
Hybrid Electric Vehicle Team of Virginia Tech

DC/DC Converter
Inverter
Air Compressor
Fuel Cell Controller

Contactor Box Charge Controller

A/C Compressor H 2 Fuel Tanks

Fuel Cell Heat Exchanger Buck Converter

Battery Pack
Inverter Heat Exchanger
Transaxle AC Induction Motor
Boost Converter Air Compressor
Fuel Cell Cooling Pump
Deionized Water Reservoir
Humidification Pump
 Hybrid Electric Vehicle Team of Virginia Tech

Energy Usage On a Fuel Cell Hybrid Vehicle

Fuel In Heat Off Board Electricity
Fan Fan Rejection Battery In
Charger
Hydrogen
Storage

Fuel Purge
Battery Traction
Air Pack Motor/
Air Compressor
Exhaust Inverter

Boost Converter
Air
In Air HVAC
DC/DC
Comp Comp HVAC
Fuel Conv
Inv Inv Load
Air
Electrical
Hybrid Electric Vehicle Team of Virginia Tech

Test data shows that the system performed well

120 Test Date: 5/19/99
Ideal
100
Minimum

Observed
Voltage (VDC)

80
voltages

60

40

20

0
0 50 100 150 200 250 300 350
Current (A)
Hybrid Electric Vehicle Team of Virginia Tech

City Cycle - Fuel Cell Efficiencies

0.8
0.7
0.6
Eff (%/100)

0.5
(Gross Stack Power)
0.4
0.3
0.2
(Net System Power) Stack Efficiency
0.1
System Efficiency
0
0 2500 5000 7500 10000 12500 15000
Power (Watts)
Hybrid Electric Vehicle Team of Virginia Tech

AD V I S OR
• Proven vehicle modeling platform for:
– Vehicle Energy Use and Emissions
– Powertrain Performance and Control
• Easy to use
– User-Friendly Graphical Interface
– Easily interchange and analyze:
• Components
• System Configurations
• Control Strategies
Hybrid Electric Vehicle Team of Virginia Tech

AD V I S OR
• Uses Matlab and Simulink
– Component building blocks
– Data files load component specifications
• Other Features Include:
– Parametric analysis
– Sensitivity studies
– Combined speed and grade traces
– It’s Free! www.ctts.nrel.gov/analysis
Hybrid Electric Vehicle Team of Virginia Tech

• EV Model + Engine and Generator

• Backward/Forward Framework
gal
vehicle controls total fuel used
(gal)
<vc>
t
Clock To Workspace ex_calc

generator/ series hybrid AND

<sdo> controller <gc> control stategy fc_emis
fuel
<cs>
converter
<fc> emis
HC, CO,
exhaust sys NOx, PM (g/s)
drive cycle <ex>
<cyc>

Version & vehicle <veh> wheel and motor/

final drive <fd> gearbox <gb>
energy
Copyright axle <wh> controller <mc> power
bus <pb> storage <ess>
Hybrid Electric Vehicle Team of Virginia Tech

• 2 Years of Hybrid Modeling Experience

– Validation of ADVISOR’s HEV capabilities
– Engine emissions mapping
• No Previous FCV Experience
– Components tested by OEM and in house
– Individual systems modeled as completed
– Building-blocks for a full-system model
Hybrid Electric Vehicle Team of Virginia Tech

• EV + Backward Looking Fuel Cell

System Model
Parasitic Loads

vehicle controls fuel_use_grams_sec

<vc> total fuel used
t
(grams/sec)
Clock To Workspace
Fuel Cell Boost Converter Fuel Cell <fc>
Control Strategy <bc>
<sdo> <cs>

emis
HC, CO,
NOx, PM (g/s)
drive cycle
<cyc>

Version & vehicle <veh> wheel and final drive <fd> gearbox <gb> motor/
Copyright axle <wh> controller <mc> power energy
bus <pb> storage <ess>
 Hybrid Electric Vehicle Team of Virginia Tech

6 0
City Driving Cycle
5 0

4 0
S P E E D (m ph)

3 0

2 0

1 0

0
0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0 1 4 0 0
T IM E (s e c )

VEHICLE VELOCITY TRACE (MPH)

55000

35000
Power (W)

15000

-5000
0 200 400 600 800 1000 1200
Time (sec)
-25000

Motor Controller and Accessory Power (vehicle data and vehicle model)
Hybrid Electric Vehicle Team of Virginia Tech

10000
Data
Model
7500
Power (W)

5000

2500

0 Time (s)

0 200 400 600 800 1000 1200 1400

Hybrid Electric Vehicle Team of Virginia Tech

City Cycle - Energy Use

Energy Sourced for use by the Drive Motor
3000
CITY
2500
MPGGE **
Model: 29.1
Energy (kW-hour)

2000
Dyno: 29.1
1500

1000 Battery
Fuel Cell
500

0
Model Dyno **SOC Corrected
 Hybrid Electric Vehicle Team of Virginia Tech

60

50

40
D(mp)
h

30
SPEE

20

10

0
0 100 200 300 400 500 600 700 800
TIM E (s e c )

VEHICLE VELOCITY TRACE (MPH)

60000

40000
Power (W)

20000

0
0 100 200 300 400 500 600 700 800
-20000
Time (sec)
-40000
Motor Controller and Accessory Power (vehicle data and vehicle model)
Hybrid Electric Vehicle Team of Virginia Tech

8500
Data Model

7000
Power (W)

5500

4000
300 320 340 360 380 400
Time (s)
Hybrid Electric Vehicle Team of Virginia Tech

Time (s)
0
-1 0 200 400 600 800

-2
Amp hours

-3
-4
-5
-6
Data
-7
Model
-8
Hybrid Electric Vehicle Team of Virginia Tech

Highway Cycle Bus Voltage Tracking

400
375
350
Voltage (V)

325
300
275
250
Dyno Batt Volts
225
Model Batt Volts
200
0 200 400 600 800
Time (s)
Hybrid Electric Vehicle Team of Virginia Tech

Highway Cycle - Energy Use

Energy Sourced for use by the Drive Motor
6000
HIGHWAY
5000 MPGGE **
Model: 28.3
Energy (kW-hour)

4000
Dyno: 26.7
3000

Battery
2000
Fuel Cell
1000

0
Model Dyno **SOC Corrected
Hybrid Electric Vehicle Team of Virginia Tech

• Accurate for....
– Trends and transients
– Energy generation and energy storage
– Identifying inefficient systems
• Lacking because
– Not easily adaptable
– No forward looking capabilites
– Inter-system impacts not considered
Hybrid Electric Vehicle Team of Virginia Tech

• Lumps Fuel Cell Accessories and Stack

1
pw r req
(W )
S um
1
Net pw r out a (W )
SUM
Auxilaries Fuel C ell S tack Auxilaries1
2
em is info
3
prev fuel use (g/s) fuel use (g/s)
Hybrid Electric Vehicle Team of Virginia Tech

• Restores Forward Looking Capability

• Eases Customization
• Stack Data Entry Options:
– Polarization Curve
– Power vs. Efficiency
– Interface with Argonne’s GCTool
• Load Following Accessories
Hybrid Electric Vehicle Team of Virginia Tech

Effect of Repaired Fuel Cell on Model Results

Energy Sourced by the Drive Motor - "Generic Model" Results
2500

2000
Energy (kW-hour)

1500 City Cycle Fuel Economy (mpgge) **

Contsant Speed Air Compressor: 30.5
1000 Air Compressor Load Following: 43.6

500 Battery
Fuel Cell
0
Model
Constant speed case VariableDyno
speed case
-500
**SOC SUSTAINING!
Hybrid Electric Vehicle Team of Virginia Tech

• Further Testing of ANIMUL H2

– Verification of as-designed performance
– Load following efficiency improvements
• Thermal Modeling
– Interaction between cooling and humidity
– Methods to speed warmup times
• Investigate Effciency and Cost vs.
Degree of Hybridization
 Hybrid Electric Vehicle Team of Virginia Tech

City Driving Cycle - Thermal Data

14000
12000
Power (W)

10000
8000
6000
4000
2000
0
0 200 400 Time
600(s) 800 1000 1200 14
Fuel Cell System Gross Power
70
60
50
Temp. (C)

40
30
20
10
0
0 200 400 600Time (sec) 800 1000 1200 1400
Fuel Cell Coolant Output Temperature
Hybrid Electric Vehicle Team of Virginia Tech

100

80

60

40
Air Temp (C)
20 Air Humidity (RH%)
Water Temp (C)
0
0 200 400 600 800 1000 1200 1400
Time (s)
Hybrid Electric Vehicle Team of Virginia Tech

• Developed accurate component models

• “Generic Model” Validated and Now
Released in ADVISOR 2.2

• NREL Systems Analysis Team:

Keith Wipke, Tony Markel, Sam Sprik
• HEVT, Dr. Doug Nelson, Energy Partners
Hybrid Electric Vehicle Team of Virginia Tech

Thanks for your interest!!

The Hybrid Vehicle for the Next Millennium

www.futuretruck.org
Hybrid Electric Vehicle Team of Virginia Tech

The Hybrid Vehicle for the Next Millennium

Additional Slides
Hybrid Electric Vehicle Team of Virginia Tech

• Fuel Infrastructure!!
• Design for Manufacturability / Cost
• Weight Reduction = Efficiency + Range
• System Performance
– Current customer expectations worldwide
– Durability over 100,000 mi or 5000 hr
• Education of the public:
If you build it, will they come?
Hybrid Electric Vehicle Team of Virginia Tech

Fuel
Heat Exchanger

Cell

Sensor tank Sensor tank

(to humidification) Pump

D/I
Bed
Water
Reservoir
Hybrid Electric Vehicle Team of Virginia Tech
Hybrid Electric Vehicle Team of Virginia Tech

Trunk Fill Port

Pressure Regulator (dome loaded)

Hydrogen Storage Vent Mass Flow

Meter

(Recirc.)

Fuel
Cell

From Cooling System

Separator
Humidification

Engine Compartment
Hybrid Electric Vehicle Team of Virginia Tech

DC/DC Converter
Contactor Box
HVAC Compressor
(Dashed) Buck Converter Air Compressor
Inverter HX Air (for Air Compressor) Motor Controller
Compressor

High Voltage Lines

Boost
Converter Inverter Cooling
Reservoir
H2 H2
HVAC Coil Battery
Traction Motor/ Fuel
Fuel
Transaxle Box Tank
Fuel Cell HX Tank (under
Inverter 12V Battery car)
(Dashed)
Fuel Cell Stack Relief Valve Manual Shutoff
(Dashed)

Vacuum
Pump
Fuel Cell Cooling Deionized Power Steering
Water Pump Water Tank
Pump (Dashed)
Humidification
Front
Humidification
Recirc. Pump Pump
Note: Components denoted by a dashed line are
located above other components
 ADVISOR and Solid Modeler Integration
Mark Henault, ESS West, Inc.

The Integration of ADAMS/Car with ADVISOR,

Realization of the "Functional Digital Vehicle”
Presentation (Joshi)
Paper (Spagnuolo)
Sandeep Joshi and Tony Spagnuolo, Mechanical Dynamics Inc.
 ADVISOR & Solid Modeler Integration

INTRODUCTION

Mark Henault
ESS West, Inc.

l Working with NREL on integration of ADVISOR with Solid

Modeling tools.
l Contact: [email protected]
l Phone: 303-429-5005

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

OVERVIEW

l Pro/HEV
– First integration of ADVISOR with Pro/Engineer

l Functional Digital Vehicle (FDV)

– Next generation of ADVISOR integration with solid modeling
& other tools (Pro/E, CATIA, Dyna, ADAMS, etc.)
– Currently underway at NREL

l Integration of ADVISOR with Solid Modeling Tools

– Why integrate
n Packaging
n Occupant Comfort
n Crash/Structural Analysis
– Current Status
– Future Plans

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

What is Required for a FDV?

l Power-train optimization (ADVISOR)

– Fuel economy
– Emissions
– Acceleration & driving performance

l Vehicle dynamic optimization (ADAMS/Car)

– Handling/performance prediction
– Durability prediction
– Vehicle stability

l Packaging – (1st order performed by Pro/HEV)

– Under hood
– Trunk
– Under vehicle
– Vehicle center of gravity, air volumes, air flow studies, etc.

l Occupant comfort (interior volume provided by Pro/HEV)

– Significant energy consumption for A/C & heating
– Minimize energy while maximizing Occupant comfort
– Not merely function of interior air temp & velocity of air

l Crash & Structural performance (structure provided by FDV)

– Occupants safe?
– 5mph vehicle damage?

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Pro/HEV & FDV

l Can recall “known” vehicle, modify “known” vehicle,
or create a new vehicle from scratch in Pro/HEV.
– Vehicle center of gravity
– Vehicle mass
– Packaging
n Pro/HEV stores 250 dimensions to control vehicle
configuration
n Contains pre-defined suspension configurations to integrate
with ADAMS (MacPherson in Pro/HEV, more planned for
FDV).

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Pro/HEV & FDV

l Capture interior for CFD Analysis

– Assigns boundary conditions for glass, etc.
– Default conditions can be solved quickly

l Vehicle morph, generation of vrml files, e-mail

notification, download files, visualize files
– 5 minutes plus download time

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Loading ADVISOR Vehicle into Pro/HEV

ADVISOR User Conference August 25, 2000

ADVISOR & Solid Modeler Integration

“What If” Studies

ADVISOR User Conference August 25, 2000

ADVISOR & Solid Modeler Integration

“What If” Studies

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

E-mail notification of VRML files

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Visualizing VRML Vehicle

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Battery Packaging Comparison

Prius (NiMH) Battery “A” (Li-ion)

Optima (Pb-Acid)

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Why Worry About Aux Loads?

1 mpg => $4B annually

1X Vehicle (3100 lb., 3.0-L, SI, 800 W

90 acc., 26.8 mpg comb.)
3X Hybrid (2000 lb., 1.3-L, CIDI,
80
SCO3 Fuel Economy (mpg)

Parallel, 400 W acc., 81.5 mpg comb.)

70

60

50

40

30

20

10

0
0 1000 2000 3000 4000
Accessory Load (W)

ADVISOR User Conference August 25, 2000

ADVISOR & Solid Modeler Integration

The Modeling Process

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Five Step Analysis Process

1. Define interior air volume (CAD File)

2. Create mesh & assign thermal B.C. (ICEM-CFD)

3. Perform thermal/fluid analysis

4. Calculate transient thermal occupant comfort

5. Calculate fuel economy, and tail pipe emissions for

specified drive cycles (ADVISOR)

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Step 1 – Define Interior Air Volume

l Traditionally get IGES data (partial) from OEM.
l Spend weeks cleaning up file to generate usable volume.
l Hope no major changes take place while working on the file.

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Step 1 – Using the FDV Approach

l Begin with Parametric

vehicle.
l Morph vehicle to match
given vehicle’s interior
(automatic for pre-defined
vehicles)
– Registers
– Surfaces
– Glass area

l Package Parametric Seats &

Occupants (next slide).
l Generate Volume.

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Step 1 - Digital Occupants

l Integrated digital manikin
– links CFD and
thermal comfort model
– experimental validation
of thermal comfort
manikin

ADVISOR User Conference August 25, 2000

ADVISOR & Solid Modeler Integration

Step 1 – “Morphed” FDV

ADVISOR User Conference August 25, 2000

ADVISOR & Solid Modeler Integration

Step 1 - Interior Volume

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Step 2 - Mesh & Boundaries

l ICEM-CFD

– Applies B.C. to vents,

glass, body, etc.

– Generates mesh of
volume

– Writes input file for

analysis codes

– Automatically
launches thermal/fluid
solver

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

l Objective: To specify the cabin thermal properties and

boundary conditions.

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Step 3 - Perform Thermal/Fluid Analysis

l FLUENT, STAR/CD simulates thermal/fluid

environment

l ICEM-CFD retrieves and visualizes results from

thermal/fluid simulation

l ICEM-CFD generates input files for Occupant

Thermal Comfort Model

l Interior Cool-down Animation

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Results - 0, 10, and 20 minutes

T=0 min

T=10 min

T=20 min

Standard Windshield – IR Reflective All Glass – IR Reflective

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Step 4 – Thermal Comfort Prediction

l Thermal Comfort Model

– Multi-segmented model
– Clothing w/moisture and thermal
capacitance
– Data every 30 sec from ICEM-CFD for
20 min simulation
– Transient solution
– Advanced control strategies
– Maximize comfort to thermally sensitive
areas vs. uniformity

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Step 5 – Vehicle Simulation, ADVISOR

l Fuel economy prediction for drive cycle

l Tailpipe estimate
l Transient A/C model w/boiling
l A/C size and performance for drive cycle

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Summary
l Brief Overview of the benefits of ADVISOR &
Solid Modeling Integration
– Packaging studies
– Mass & CG calculations
– Interior volume generation for occupant comfort
analysis & a/c sizing
– Generation of geometry for future work (ADAMS,
Crash analysis, etc.)

ADVISOR User Conference August 25, 2000

 ADVISOR & Solid Modeler Integration

Future Direction: Functional Digital Vehicle

ADVISOR User Conference August 25, 2000

 Sandeep Joshi
Mechanical Dynamics, Inc.

Users Conference for ADVISOR

Costa Mesa, CA
August 24-25, 2000
Presentation Overview

■ Introduction to Mechanical Dynamics, Inc. (MDI)

■ Introduction to ADAMS/Car
■ Interfacing ADVISOR with ADAMS/Car
■ Closing Remarks

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Presentation Overview

■ Introduction to Mechanical Dynamics, Inc. (MDI)

■ Introduction to ADAMS/Car
■ Interfacing ADVISOR with ADAMS/Car
■ Closing Remarks

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Introduction to MDI
■ Founded in 1977
■ Pioneered the field of Mechanical System Simulation.
■ ADAMS line of software products is the most widely
used MSS software in the world
■ Vehicle dynamics, general dynamics, controls, space
structures, mechanism modeling, aerospace systems
simulations, railcar and motor-sports applications
■ Ten offices in the United States and wholly owned
subsidiaries in several countries around the globe
■ http://www.adams.com

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Functional Virtual Prototyping
■ Functional simulation of real-world systems
■ Reduces time-intensive, costly, hardware build-test-
refine cycles
■ Improves quality (cut warranty and
repair/maintenance costs)
■ Managing risks through better information upfront
■ “what-if” scenarios

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Presentation Overview

■ Introduction to Mechanical Dynamics, Inc. (MDI)

■ Introduction to ADAMS/Car
■ Interfacing ADVISOR with ADAMS/Car
■ Closing Remarks

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Introduction to ADAMS/Car
■ Consortium of automakers including Audi, BMW,
Renault, and Volvo
■ Fully parametric
■ Build and simulate virtual prototypes of vehicles and
vehicle assemblies
■ Software Modules: Suspension Design, Conceptual
Suspension, Vehicle Dynamics
■ Add-on Capabilities: Engine Simulation, Driveline
Simulation, Tire Simulation, Driver Simulation

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Introduction to ADAMS/Car
■ Interfaces seamlessly with other ADAMS products
(including ADAMS/Insight for DOE)
■ Demo ADAMS/Car
◆ Two operational modes, Template Builder and Standard Interface
◆ Kinematic and/or Compliant Subsystems
◆ Flexible Bodies
◆ Data Libraries
◆ Curve Manager
◆ Load Cases
◆ Plot Configuration

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Presentation Overview

■ Introduction to Mechanical Dynamics, Inc. (MDI)

■ Introduction to ADAMS/Car
■ Interfacing ADVISOR with ADAMS/Car
■ Closing Remarks

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Outline of Interface

■ Two interface approaches will be used:

◆ ADAMS/Car - ADVISOR Cosimulation

◆ Export to ADAMS/Car

◆ Each approach has its own advantages and serves different

simulation purposes.

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

ADAMS/Car - ADVISOR Cosimulation
■ Overview
◆ Linking ADAMS/Car full vehicle model with ADVISOR model
◆ Both ADAMS and Simulink solvers run together
◆ Information passed back and forth between the two at each time
step

■ Method
◆ ADAMS/Car full vehicle model using customized powertrain
template
◆ Use the ADAMS/Controls interface for exporting the plant
◆ Modified ADVISOR model to work with ADAMS/Car model

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

ADAMS/Car - ADVISOR Cosimulation
■ General Benefits
◆ Share Models

Design Process before ADAMS/Controls Improved Design Process after ADAMS/Controls

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

ADAMS/Car - ADVISOR Cosimulation
■ General Benefits (contd.)
◆ Verify combined effects of a control system on a non-linear non-
rigid model
◆ Eliminate risk of having control law developed for the wrong
mechanical model
◆ Add sophisticated controls to your ADAMS model
◆ Generate mechanical system simulation models directly from your
ADAMS data
◆ Analyze results of the cosimulation either in the ADAMS or the
control application environment

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

ADAMS/Car - ADVISOR Cosimulation
■ Specific Benefits
◆ Simulate 4WD/AWD powertrains
• Torque split can be actively controlled by ADVISOR
◆ Calculate energy losses during handling/durability events
• Useful for trying minimizing losses for maximum fuel efficiency
◆ Integrate accessory loads (like electric power steering) and look at
their energy impact vs performance
◆ Trade-offs to accurately assess impact of vehicle/component mass
reduction and evaluating effect on dynamic performance

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

ADAMS/Car - ADVISOR Cosimulation

Preprocessing Analysis Postprocessing

Pro/Engineer

ADAMS/Car
Full Vehicle Model
Mass, Inertia Handling Results,
& Geometry Animations

Co-simulation

ADVISOR Energy losses, Emissions,

Fuel efficiency, etc.

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

ADAMS/Car - ADVISOR Cosimulation
■ Information Flow
◆ The major variables exchanged are shown below. Additional
information is also exchanged.
ADAMS/Car
Full Vehicle Model

Vehicle Dynamics &

Stability Analysis

Torque at Wheels Cosimulation Vehicle Velocity

ADVISOR

Hybrid powertrains,
Energy management,
Controls

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

ADAMS/Car - ADVISOR Cosimulation
■ Exporting ADAMS/Car Plant to Simulink

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

ADAMS/Car - ADVISOR Cosimulation
■ Demo Cosimulation
◆ ADAMS Plant Mask

■ Results can be read back into ADAMS for

postprocessing (animation and plotting)
Users Conference for ADVISOR, Costa Mesa, CA, August 2000
Export to ADAMS/Car
■ Overview
◆ ADAMS/Car full vehicle model with mass and inertia properties
exported from ADVISOR
◆ One way information flow to ADAMS/Car

■ Method
◆ Output mass and inertia properties from ADVISOR to ADAMS/Car
• Optionally, geometry may be specified in web interface
◆ Run standard handling maneuvers in ADAMS/Car

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Export to ADAMS/Car
■ Benefits
◆ Faster simulations
◆ Quick estimate of handling performance of hybrid vehicle
• Example: allows analysis of battery pack location (often a large
mass) and effect on handling

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Export to ADAMS/Car
■ Information Flow

ADVISOR Digital Functional

Web Interface

ADAMS/Car Property File

ADAMS/Car Full Vehicle Model

Results for Handling Events

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Presentation Overview

■ Introduction to Mechanical Dynamics, Inc. (MDI)

■ Introduction to ADAMS/Car
■ Interfacing ADVISOR with ADAMS/Car
■ Closing Remarks

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Status
■ Work in progress
■ ADAMS/Car - ADVISOR Cosimulation
◆ Proof of concept phase completed.
◆ Need to modify ADVISOR to be used in the cosimulation mode

■ Export to ADAMS/Car
◆ Interface for reading in an ADAMS/Car property file into an
ADAMS/Car full vehicle model completed.
◆ Need to develop interface to write out these property files from
within ADVISOR (or any other source)

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

Conclusions
■ ADAMS is the world’s most widely used mechanical
system simulation tool.
■ ADAMS is the de facto standard for functional
system-level virtual prototyping in the automotive
industry.
■ Cosimulation is a proven technology that allows us to
study the effects of a control system on a non-linear
non-rigid mechanical system model
■ As applied to this interface, cosimulation will allow us
to simulate 4WD/AWD conditions, study energy
losses in handling events, and include accessory
loads in the model
Users Conference for ADVISOR, Costa Mesa, CA, August 2000
Conclusions (contd.)
■ The “Export to ADAMS/Car” method will facilitate
getting quick estimates on the handling behavior of
hybrid vehicles

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

The Team
■ MDI
◆ Chris Gaddy
◆ Sandeep Joshi
◆ Rachael Kadell
◆ Matthew Ma
◆ Jennifer Peeples
◆ Tony Spagnuolo

■ NREL
◆ Aaron Brooker
◆ Terry Penney
◆ Sam Sprik
◆ Keith Wipke
Users Conference for ADVISOR, Costa Mesa, CA, August 2000
Feedback
■ Comments or questions?
■ Contact Information
◆ [email protected]
◆ 949-786-0600 x100

Users Conference for ADVISOR, Costa Mesa, CA, August 2000

 The Integration of ADAMS/Car with ADVISOR,
Realization of the "Functional Digital Vehicle”
Tony Spagnuolo
Mechanical Dynamics Inc.

ABSTRACT during the development process, or risk program

cancellation.
As product complexity increases with hybrid electric
vehicles (HEV) and competitive product development The question is how does the automotive industry today
cycle times are reduced, hardware prototype creation deliver vehicles that meet customer expectations? If we
and testing becomes one bottleneck in successful HEV look at one successful vehicle, the current BMW 3-series
launches. Due to this bottleneck, leading global sedan came to market amidst a flurry of accolades and
manufacturers are feeling increasing pressure to rapidly awards. “Perfection down to the last detail” was an
institute enterprise-wide, simulation-based design and overriding philosophy throughout the design process
virtual prototyping practices that can insure greater used to create this latest version of “the Ultimate Driving
product performance and quality in a fraction of both the Machine.” According to BMW Magazine [1], the
time and cost required with traditional build-and-test development process involved five and a half years, 2.6
approaches. This paper outlines the functional virtual million man-hours, 130 hand-made system-level
prototyping implementations for HEVs and discusses the hardware prototypes created at a cost of roughly
industry trends supporting such a shift. Specifically a $350,000 per vehicle.
connection between ADVISOR and ADAMS is outlined
and discussed. Assuming BMW has a competitive and representative
vehicle development process, it is unlikely that a single
HEV program will be able to afford such a process.
However, without such a process it is likely that an HEV
development program will not meet market standards.
How can HEV program increase its confidence that it will
meet market demands without the high cost of a physical
prototype based development process? The answer is
the Virtual Prototyping.

Simulation-based design practices allow product

designers, engineers, and analysts to more quickly
assess form, fit, function, and manufacturability of new
products from concept design to production. No longer
is it necessary to wait months to build a hardware
prototype to make a small number of expensive
modifications in order to assess proposed design
changes. Instead, participants in the design process are
Figure 1. Moving from Physical to Virtual Prototyping able to construct accurate virtual prototypes, exercise
the models through hundreds of tests with thousands of
variations, and optimize the form, fit, function, and
INTRODUCTION manufacturing characteristics in a fraction of the cost
and time of traditional hardware prototype processes.
There is a growing emphasis to develop hybrid electric T h e V i s io n
M o v e a w a y fro m :
vehicles (HEV), since they offer the opportunity to C oncept P ro d u c t
reduce energy consumption and emissions while still
providing the consumer the utility of today’s passenger
vehicle. There are great engineering challenges in
developing the practical power plants for hybrid electric
vehicles; however, the ultimate challenge is consumer D e s ig n - B u ild - M ea s u re

acceptance. Consumers have ever-increasing demands T o w a rd s : V irtu a l P ro to ty p e

on ride, handling, durability and convenience features. Concept P ro d u c t

Without a dramatic shift in market dynamics (e.g. a

dramatic increase in fuel costs), it is unlikely that
consumers will sacrifice their demands when selecting a D e v e lo p m e n t t i m e W IT H v ir t u a l p r o t o t y p e s
T im e = $

vehicle in the future. Therefore, an HEV concept must D e v e lo p m e n t t i m e W IT H O U T v ir t u a l p r o t o t y p e s

demonstrate its ability to meet market requirements

Figure 2. Moving from Physical to Virtual Prototyping
TRADITIONAL CAD/CAM/CAE VS. SYSTEM-LEVEL assembled product (Functional Virtual Prototyping), not
VIRTUAL PROTOTYPING just the component parts. And they need to investigate
the entire manufacturing and assembly of the product
Traditional CAD/CAM/CAE practices throughout the (Virtual Factory Simulation), not just the creation of the
1970’s and 1980’s focused on a concept referred to as parts. As global product manufacturers began to realize
“art-to-part.” Most engineering software activity was this fact over the last 2-5 years, it was natural for them to
oriented toward the design, development, and look for extensions to their traditional CAD/CAM/CAE
manufacturing of higher quality parts. Detailed, three- systems to address system-level design. Part-focused
dimensional solid modelers (CAD) allowed for quick part CAD/CAM/CAE providers hurried to extend their
design and understanding of “form.” Finite element software to address system-level designs with varying
software (CAE) made it feasible to perform detailed levels of success. But simple extensions of part design
meshing and analysis of structural effects, thermal paradigms to system-level design often lead to
effects, and vibratory characteristics, or “function,” of impractical software products. For instance, as
individual parts. Software aimed at improving designers and engineers tried to construct large
“manufacturability” of parts (CAM) provided better assemblies of solid models to facilitate system-level
control of machine tools, robots, mold procedures, interference detection and virtual fly-through, the
stamping procedures, forging processes, etc. rendering performance of most traditional
CAD/CAM/CAE systems became unacceptably slow
These traditional CAD/CAM/CAE tools and processes (e.g., measured in hours). Similarly, engineers and
were embraced and implemented throughout major analysts investigating system level operating
industries, including the automotive, aerospace, general performance attempted to combine all of their
machinery, and electromechanical markets. For the component finite element models and perform nonlinear
most part, they lived up to their promise of dramatically finite element system simulations. These typically took
improving part design. In the automotive industry, for Cray-weeks of simulation time to predict only seconds of
example, automotive suppliers reported a 40% reduction real operating performance, thus making design trade-off
in part defects over a recent five-year period. This
significant improvement was accompanied by a System-focused Virtual Prototyping
corresponding drop in development and manufacturing
costs attained through successful implementation of Virtual Virtual Prototyping & Virtual
better CAD/CAM/CAE tools and processes. Mock-up Design Refinement Production

Unfortunately, during the same five-year period that

automotive part suppliers were achieving a 40% Motion/
reduction in part defects, the vehicle manufacturers System
Handling Tolerance
Assembly
(OEMs) who were using these parts to assemble and
Fly Vibration/
market full vehicles experienced only a 20% reduction in Noise Robotics
Through
warranty costs. In some sense, this was a surprise to
Motion Durability/
many OEMs who expected a one-to-one Fatigue Assembly
Envelopes
correspondence between part defects and warranty
Safety/
costs. In retrospect, it seems perfectly sensible. Collisions Impact Sequencing
Optimal part design rarely leads to optimal system
design. For example, when perfectly good brakes are ... Emissions
Fuel Usage
...
combined with a perfectly good suspension system and
a fine chassis, the resulting combination often performs
...
in a less-than-stellar manner. Clearly, the interaction of investigations impractical. Similar problems occurred in
form, fit, function, and assembly of all parts in a product manufacturing and assembly.
Figure 3. Technology Segments of System-Focused CAD/CAM/CAE
is a major contributor to overall product quality. We may
be reaching levels of diminishing return in applying
CAD/CAM/CAE technologies to part design. The big
opportunity to increase quality and reduce time and
cost has now shifted to the system level. New methodologies, specifically oriented toward rapid
system-level design, had to be adopted. The growth in
More significant returns on investment can be realized simulation-based design tools has now shifted away
today through the effective use of simulation-based from traditional CAD/CAM/CAE software and toward
these newer system-focused solutions. Specifically,
design processes and virtual prototyping applied to
these system-level solutions include Digital Mock-Up
system-level design. Manufacturers now need a
tools to investigate product form and fit, Functional
means to quickly assess form and fit of entire
Virtual Prototyping tools to assess product function and
assemblies of three-dimensional solid models
operating performance, and Virtual Factory Simulation to
comprising a product (Digital Mock-Up). They need to
assess manufacturability and assembly of the product.
be able to assess the operating function of the entire
Enterprise-wide, Product Data Management (PDM) is Deployment of Functional Virtual Prototyping typically
the “glue” that enables these system-focused solutions involves five phases: Build, Test, Validate, Refine, and
to be successful by making all of the up-to-date Automate.
component data readily available and manageable.

Digital Mock-Up (DMU) solutions that make efficient use

of tessellated three-dimensional component solid models
System-focused Virtual
were pioneered by Tecoplan, Engineering Animation, P t t i
Clarus, and Division among others. These allow efficient Build Test Validate Refine Automate
design collaboration, mark-up, fly-through, and
interference/collision detection. Integrated with Product
Data Management Systems, these Digital Mock-Up
products provide an excellent means to insure that all of
the parts of the product will fit together properly and that
the product will appear as specified. Product Data Management System

Functional Virtual Prototyping solutions make efficient

Figure 4. Five Phases of Functional Virtual Prototyping
use of three-dimensional component solid models and
modal representations of component finite element
models to accurately predict the operating performance BUILD
of the product in virtual lab tests and virtual field tests.
Mechanical Dynamics pioneered this field with its During the Build phase, virtual prototypes are created of
ADAMS system simulation product line and is expanding both the new product concept and any target products
its coverage through its partnership with MTS systems, which may already exist in the market. In the early
nCode, and the solid modeling and finite element concept stage, the virtual prototype models of the new
solution vendors. product concept are kept simple and are most often
driven by desired functionality data curves, rather than
Tecnomatix and Deneb pioneered virtual Factory by specific product topologies. Appropriate target setting
Simulation. With these solutions, the entire is, of course, very important. For example, the desired
manufacturing and assembly of products can be functionality data curves should be derived from a
simulated, and field maintenance of products can be customer Quality Function Deployment (QFD) study that
assessed as well. identifies the desired operating performance. For
instance, in the initial design of a vehicle suspension
The combination of Digital Mock-up, Functional Virtual system, the virtual prototype model often involves only
Prototyping, and Virtual Factory Simulation provide a the overall vehicle body and a set of vehicle suspension
means for realizing an effective transition from hardware curves that relate the movement of the body to the
prototyping practices to software prototyping practices movement of the wheels. These data curves embody
with all of the concomitant benefits. The remainder of the desired suspension characteristics. During later
this paper will focus on the subject of Functional Virtual model refinement, specific suspension topologies are
Prototyping and the implementation of Functional Virtual chosen (e.g., McPherson Strut) and the software
Prototyping for HEVs. optimizes suspension geometry and structural properties
to yield the relationship described by the chosen curves.
FUNCTIONAL VIRTUAL PROTOTYPING Similarly, to create models of target products, the actual
target product is physically tested and its characteristics
Effective Functional Virtual Prototyping (FVP) allows the are accurately measured. This data is incorporated into
full operation of the product to be considered and a system-level model of the competitive vehicle to use
evaluated early enough in the design process to allow later during the evaluation phases.
for “function” to truly drive “form” and “fit.” It also allows
multi-function optimization to be realized, such that a
true balance can be obtained between competing
functional requirements involving performance, safety,
durability, cost, comfort, emissions, fuel economy etc.
These two benefits were largely impractical in traditional
development cycles involving extensive reliance on
hardware prototypes. In addition to these benefits,
functional virtual prototyping has proven effective in
facilitating tighter and more successful relationships
between manufacturers and their lead suppliers.
 Target Setting

QFD

Target Vehicle
Vehicle Targets Virtual Prototype
Comparison Vehicle

Sub-
System
Targets

Sub-System
Virtual Prototype

Component Targets
Component Virtual Prototype

Figure 6. Field and Lab Testing: Vritual and Physical

Effective implementations of Functional Virtual

Prototyping require a tight synergy between physical
testing of hardware prototypes (components and
Figure 5. Target Setting via QFD
systems) as well as simulation-based testing of virtual
A modular system design process facilitates functional prototypes (components and systems). Testing
virtual prototyping and the manufacturer-supplier requirements vary during the different stages of the
interaction. Clear inputs and outputs between various design process. At the outset of a new product design
subsystems permit the development of multiple based on virtual prototyping, hardware testing is
subsystem models with varying levels of model fidelity instrumental in two ways. First, component tests are
and complexity. These subsystem and system-level performed using various real component alternatives.
virtual prototypes are comprised of rigid and flexible These tests provide good characteristic data for a
representations of component parts connected through complete system-level virtual prototype model.
mathematically defined constraints. The geometry and Secondly, full system hardware tests are conducted
mass properties for the parts are derived from using target products. This allows for the simultaneous
component solid models; while the structural, thermal, development of virtual prototypes of competing products
and vibratory properties are derived from component so that performance comparisons can be made
finite element models or experimental tests. The most throughout the design and development cycle.
effective implementations of virtual prototyping begin in
this Build phase with a close cooperation between Then, during concept design, virtual testing is used to
engineering analysts and test engineers. Also, up-front exercise the new system model through a limited
planning of the product parameters that may be varied in number of actual test scenarios such that performance
the design cycle and how manufacturers and suppliers data can be collected and validation can be performed.
are going to share models, can be tremendously helpful. For companies that are initiating new virtual prototyping
processes, it is imperative that they build a first system-
TEST level physical prototype at this stage in order to insure
confidence in the simulation model. Companies who
Perhaps the single most important axiom for successful have been through this process a number of times have
functional virtual prototyping is to simulate as you test. learned how to validate the modeling assumptions such
Testing of hardware prototypes has traditionally involved that a physical prototype is unnecessary at this stage.
both lab tests and field tests in various configurations.
With virtual prototyping, we need to create virtual Once initial validation has been achieved by correlating
equivalents of the lab tests and the field tests. By doing the test results of the physical and virtual system
this, we greatly facilitate model validation through prototypes, the true value of virtual prototyping begins to
testing, and we break down the cultural barriers to the become apparent. Thousands of system variations,
adoption of virtual prototyping practices. With regard to component choices, parameter choices, and tolerances
lab tests, successful virtual prototyping dictates that we can be examined through simulation and the results can
need to construct virtual test rigs that reproduce the test be used to confidently make design choices about the
procedures and boundary conditions of the real fixture new product. This will be discussed later in the section
and machine. With field tests, we need to construct entitled “Refine.”
models that represent the actual operating conditions of
the product in the field. This will involve virtual test tracks
in automotive.
Testing remains an important part of functional virtual In a typical validation process, the physical and virtual
prototyping throughout the design cycle. Virtual testing models are tested identically and baseline results are
is conducted continuously. Physical testing is introduced derived. The results are compared either manually or in
at various stages to either re-validate the model after automated computer-based fashion. Discrepancies are
significant refinement or to test certain configurations of noted in specific performance results. Design sensitivity
the product containing design parameters outside of analyses are performed on the virtual model to identify
those for which the model has been validated. design parameters or model areas that significantly
contribute to the performance results that do not
VALIDATE correlate. Then, a mixture of manual changes and
computerized nonlinear optimization techniques are
The importance of accurate validation of system-level employed to make changes to the model parameters
models and modeling assumptions should not be under- identified or the test procedures until acceptable
emphasized. Functional Virtual Prototyping can yield a correlation is achieved and the model is validated across
wealth of information to support rapid decision-making. the different tests.
It is critical to insure that this information reflects the
actual operating performance of the new product. The With experience, modeling assumptions can be
validation phase is not overly difficult, but often is not correlated and catalogued. This allows for the
approached with as much rigor as is warranted. The automated creation of new product virtual prototypes
companies with the very best records in making effective that can be utilized with confidence without the need for
use of Functional Virtual Prototyping have invested the construction and testing of an initial physical
significant time and resources in building a validation prototype. Physical prototypes are still needed
library. This library defines how models need to be downstream in the process to verify the design prior to
constructed so that simulation performance results can production.
be easily compared with test results. The library
catalogues past validation work and summarizes REFINE
modeling assumptions that have been validated. And
the library is integrated with an internal Product Data Refining a virtual prototype involves two aspects, refining
Management (PDM) system so that data and information the fidelity and breadth of the model, and refining the
is readily available. product design itself. Each of these will be discussed
separately here.
Good simulation tools and processes can greatly
facilitate the validation process. For instance, a As the design process progresses, the virtual prototype
simulation software product that provides quick models will be relied upon to investigate more and more
information on design sensitivity to various parameter functionality. Initially, it may be enough to understand
changes can pinpoint areas of a model to be speed of operation, the space envelope of operation, the
investigated to improve correlation between experiment total power requirements, etc. This understanding can
and simulation. Also, as stated earlier, it is important to help drive component topology selection and overall
“simulate as you test,” meaning that the same testing design parameters. Then, as issues of comfort, noise,
and instrumentation procedures should be used both in vibration, and durability need to be addressed; the virtual
the physical and virtual test process. prototype model will need to be enhanced. It is
important that the virtual prototype can access
subsystem models of varying complexity and model
fidelity. For investigations of more complex
phenomenon, it will be important to enhance the model
by replacing more and more of the rigid subsystem
models with flexible counterparts. Models of the fluid
power systems that interact with the mechanical and
electrical components will need to be represented.
Automatic control systems that alter the operating
performance of the product will need to be accurately
represented. These are all natural extensions of the
initial virtual prototype. Component and subassembly
models of varying complexity must be constructed in
such a manner so as to be quickly interchangeable. For
instance, when investigating engine performance in a
vehicle, it may be important to exercise a fairly detailed
Figure 7. Typical Parameter Study With a Virtual Prototype
engine model. However, if vehicle dynamics is the main
focus, the engine model can be effectively replaced with
a much simpler representation. A template-based
design system that allows for quick and easy exchange
of various subsystem models is of paramount
importance for effective design refinement.
 AUTOMATE (PUBLISH)

The approach outlined above leads to significantly

improved products at lower cost. To simultaneously
reduce the overall development time, it is necessary to
automate the virtual prototyping process. This phase
requires close cooperation between designers,
development engineers, analysts, and test engineers.
Automating the process can be done very effectively in
companies that make the same type of products year
after year. It is much more difficult in organizations
where radically different products are created over time.
Once the engineering analysts have worked through a
few virtual prototyping cycles and helped create
validated models that can be exercised through the
parameter changes requested by the development
engineers, the virtual prototyping environment can be
automated through the use of a template-based design
system. It works as follows. The engineering analysts
ADVISOR Flexibility Hydraulics catalogue: (1) parametric topologies that are normally
considered for new products, (2) typical parameters that
are varied in the design process, (3) the range of validity
Figure 8. Successive Refinement of a Virtual Prototype of various modeling assumptions, and (4) the different
levels of subassembly model representations required
for various levels of fidelity.
Refining the actual product design is where Functional
Virtual Prototyping delivers the real value. Once a Then an analyst utilizes a template-based design system
validated, system-level virtual prototype has been to create a series of design templates that can be used
created with interchangeable subsystem models of by the designers and development engineers to evaluate
varying model fidelity, a very rigorous design refinement design changes. These templates automate the creation
process is within reach. First, a complete battery of of the subassembly and system models. They allow
product functional tests are defined and finalized. These input only within the range of the validated modeling
will be the virtual tests used to sign-off on the new assumptions. They hide the complexity of the model by
product design. Next potential design changes are only presenting the parameter changes that have
identified in terms of component parameter changes, traditionally been varied. And they automate the
system topological changes, and potential manufacturing selection of subassembly representations in accordance
tolerances. Performing the complete battery of selected with the type of test or performance output that is
tests with all combinations of parameters and tolerances requested. If this is integrated with a PDM, it allows for
is both impractical and unnecessary. Statistics-based, quick comparisons of new design performance with
Design-Of-Experiment (DOE) methods are used to previous designs or competitive target designs. The
consider the entire universe of combinations of these analysts publish these design templates internally for
changes and determine what combinations of these use throughout the design process and even later in field
parameters must be simulated in the battery of virtual troubleshooting.
tests in order to give a statistically relevant prediction of
the envelope of operating performance. The identified This makes it possible to have an enterprise-wide virtual
combinations are then simulated using both the virtual prototyping process where any engineer in a vehicle
prototype and the battery of virtual tests, and the results manufacturing organization can access a validated
are exported to a simple spreadsheet. Curve fitting of model of any previous vehicle or current new vehicle
these results allows for quick spreadsheet assessment design. They can replace subsystems, alter vehicle
of any potential design changes within the specified design parameters, add automatic control systems, and
range. This approach facilitates rapid, knowledge-based run the vehicles through standard test procedures to
decision making in product design review meetings. understand the effects of proposed changes. This is
Requested changes to system design points or extremely powerful in stimulating creative input and
parameters can be immediately assessed for their capturing corporate design knowledge.
impact on performance, safety, durability, comfort, and
cost. Faster decisions and a better balance of
competing functional performances result from this
approach.
REALIZATION OF THE “ FUNCTIONAL DIGITAL ADAMS/Car is a family of products which includes
VEHICLE” FOR HEVS ADAMS/Pre, ADAMS/Driveline and ADAMS/Engine.
The vehicle handling and vehicle dynamics tools are
The confluence of technologies such as Digital Mock-up widely used in the automotive industry at companies
and Functional Virtual Prototyping are enabling the true such as Ford, VW, Audi and Nissan. These tools are
realization of the Functional Digital Car such that we used to investigate chassis subsystem performance,
can evaluate and optimize total vehicle performance on total vehicle handling and performance and system/
a computer. It is now possible to combine accurate subsystem loads (static and dynamic). The environment
mathematical model representations of chassis allows the inclusion of chassis subsystems, flexible
subsystems, hybrid power plant and driveline structures, control systems, drivelines, engine maps and
subsystems, and body subsystems to create a full virtual driver models. The system can start from a characteristic
vehicle. We can then simulate the performance of this based model and work up to a detailed system model
total vehicle in a virtual test lab environment or on a including full component flexibility.
virtual test track and replicate real-world behavior.

This vision is now being realized with the integration of

Full
ADVISOR with ADAMS/Car, a product of Mechanical
Dynamics Inc. (MDI). ADVISOR (Advanced Vehicle
Test Road
SimulatOR) was created by National Renewable Energy Rigs Driver
Lab (NREL) in 1994 to support the DOE (Department of
Energy) Hybrid Program. ADVISOR, created in the Chassi Engine Driveli Body
MATLAB/Simulink environment, simulates conventional,
electric or hybrid electric vehicles during driving cycles to
determine the impact on emissions and fuel economy. Suspe Valvet Trans Body-
ADVISOR is a template driven environment where Steeri
nsion Crank
rain Clutch
missio in-
validated models can be altered to determine the Brake
ng Chain/
train Differ
Tires Acc. n
Axles/ white
Restr
sensitivity of design to changes, a new design analyzed s Belt ential
or a design optimized towards specific goals. It has been Drives CV aints
extensively used by many organizations around the
world, with over 1700 copies downloaded from the Packaging Ride& Loads- Mass Vibrations
world-wide-web. Handling Reduction

Figure 10. ADAMS/Car Application Area

Figure 8. An ADVISOR Template

Figure 11. Representative ADAMS Vehicle Handling Output [2]

Like ADVISOR, ADAMS/Car is a template driven

environment. Each component of a model, including the
tests and physical data, is treated as separate template.
This allows a user to quickly “swap” one subsystem for
another, share models with a colleague or a supplier, run
design sensitivity or perform what -if studies. Further,
the modular structure eliminates the need to recreate
subsystems in new models; a previously created
Figure 9. Output Results from ADVISOR subsystem can easily be retrieved and included. The
template-based structure provides the opportunity to
shift model building from an art to production, which is a a powertrain template will be customized for easy
key requirement for Functional Virtual Prototyping to inclusion of an ADVISOR model during a simulation.
have an impact on the automotive development process.
ADAMS/Car
Full Vehicle Model
Vehicle Dyn. &
Stability Analysis

Wheel Torque
ADAMS/Controls Vehicle Speed
Co-simulation
ADVISOR

Hybrid Powertrains
Energy
Management
Controls

Figure 14: Advisor & ADAMS/Caro-Simulation

With the ADAMS/Car and ADVISOR models coupled,

there are various systems, subsystems and attributes
that can be more completely modeled and studied:

Figure 12: The ADAMS/CAR Application § 4WD/AWD powertrains (torque split actively
controlled by ADVISOR)
§ Vehicle handling/dynamics with new mass
distribution
Test
Rigs
Full Road
Driver
§
§
Stability issues related to battery placement
Determine energy losses during handling
and durability events
Chassis Power Plant Driveline Body § Determine accessory load energy impact vs.
performance
§ Asses impact of vehicle/component mass
reduction on dynamic performance
Steeri Clutc Fram
ng
Tires Acc. h e EXPORT TO ADAMS/CAR

Drive In this mode, there is one-way transfer of information

s from ADVISOR to ADAMS/Car. Since the ADVISOR
contains information (mass and inertia) that ADAMS/Car
requires, an export utility provides an easy method to
populate an ADAMS/Car model, while minimizing the
Figure 13: The ADAMS/CAR Modular Template Structure potential for data entry errors.

In a typical design process scenario, ADVISOR can be ADVISOR Digital Functional

Web Interface
used to optimized the power plant for fuel economy and
emissions. Independently, ADAMS/Car can be used to
optimize the chassis design for performance and
ADAMS/Car Property File
loading. Then the two models can be combined
together to study complete system performance.
Combining the models will have two options: Co-
simulation and export to ADAMS/Car. ADAMS/Car Full Vehicle Model

CO-SIMULATION

Co-simulation is accomplished via an interface program Results for Handling Events

called ADAMS/Controls. ADAMS/Controls links the
ADVISOR model to the ADAMS/Car model by passing
the relevant system states between the two programs
during execution at each time step. Within ADAMS/Car Figure 15: Export to ADAMS Flow
Once data is exported from ADVISOR to ADAMS/Car, it now become inexpensive and plentiful. Also, Product
is possible to get a quick determination of the vehicle Data Management systems facilitate system-level design
handling properties of a proposed HEV. For example, by making vast quantities of data available and current.
battery pack constitute a large percentage of the mass These four factors make it possible to effectively deploy
on some HEVs, so battery positioning is a major DMU and FVP today.
contributor to vehicle handling performance.
A few limiting factors still exist which retard progress in
The integration of ADVISOR is currently under applying these newer technologies. First, very few
development and scheduled to be completed in late universities have instituted effective training in these
2000. When completed, the complete system can be technologies, thus limiting the number of knowledgeable
evaluated across multiple attributes by replicating candidates for deployment. Secondly, hardware testing
standard test suites and modifying typical vehicle design is ingrained in most manufacturing organizations and
parameters throughout their acceptable ranges. Results this newer technology is sometimes viewed as a threat
can be shared globally among design teams over the rather than being synergistic. And lastly, effective
World Wide Web by means of Design-of-Experiment deployment requires some process change within these
response surfaces, plotted results, and performance large organizations and that requires a significant
animations. amount of training and the passage of time for overall
adoption.
This, in essence, is the technology that will allow vehicle
manufacturers to realize the projected dramatic
reductions in cycle times while maintaining and CONCLUSION
increasing vehicle performance, safety, longevity, fuel
economy and emissions. Effective implementation and One of the bottlenecks in developing consumer accepted
automation of functional virtual prototyping can provide a hybrid electric vehicles is the creation, instrumentation,
significant competitive edge in the market. testing, and modification of system-level hardware
prototypes. Traditional CAD/CAM/CAE methodologies
MANAGING RISK MORE EFFECTIVELY do not provide a good means to break this bottleneck.
New products in the Digital Mock-Up (DMU) area,
The result of Functional Virtual Prototyping is that Functional Virtual Prototyping (FVP) area, and Virtual
manufacturers are much better equipped to manage the Factory Simulation (VFS) provide system-level
risks inherent in the product development cycle. counterparts to traditional component-focused
Traditionally, the amount of information concerning the CAD/CAM/CAE solutions and allow for breakthroughs in
actual performance of a new vehicle was fairly low speed, cost, and quality for new product design. Key
throughout the vehicle development process until the enablers are present in the market to make these
prototype and assembly stage. Then, behavioral technologies practical today.
information increased, and risk could be reduced
through effective design changes. Unfortunately, late This paper provides a brief overview of Functional Virtual
cycle changes are very expensive and error prone. With Prototyping and how it can be successfully implemented
Functional Virtual Prototyping, behavioral performance in HEV programs. Clearly the need for this technology
predictions are obtained much earlier in the design exists. Rapidly increasing product complexity coupled
cycle, thereby allowing more effective and cost efficient with declining development budgets and time-to-market
design changes and reducing overall risk substantially. pressures mandate an alternative to singular reliance on
Reducing risk in this way has multiple benefits. It leads hardware prototype testing. New computer hardware
to vastly improved designs, limits warranty and liability and software have enabled cost-effective
issues, reduces late cycle changes and costs, and helps implementations of this FVP technology.
to reduce overall development time.
Critical success factors for FVP implementation include:
TECHNOLOGY ENABLERS AND LIMITING FACTORS
• A well-defined process
An often-asked question is ‘why haven’t technologies • System-level focus
such as Digital Mock-Up (DMU) and Functional Virtual • Effective target setting
Prototyping (FVP) been applied extensively before now?’ • Rapid simulation turnaround
To understand this, it is important to look at factors that • High quality CAE infrastructure
enable this technology and factors that inhibit it. Key
enablers include the fact that three-dimensional solid Implementation of Functional Virtual Prototyping on an
models and component finite element models are now enterprise level requires a significant commitment of
available for most system components, unlike in the time and financial resources. The benefits of making
past. Secondly, new technologies have been developed this commitment are enormous in terms of return-on-
for simplifying the representation of component data so investment and global competitiveness.
that it can be efficiently processed in large system
simulations. Thirdly, fast graphic workstations that can
quickly analyze and display system-level models have
ACKNOWLEDGMENTS

This paper was a cumulation of ongoing work and work

performed over several years at Mechanical Dynamics
and National Renewable Energy Lab (NREL). From
Mechanical Dynamics Inc., the following people
contributed to this paper:

Robert Ryan, President

Rachael Kadell, Manager, Western Region Consulting
Sandeep Joshi, Sr. Engineering Analyst

From NREL's Vehicle System Analysis Team:

Terry Penny, Technology Manager

Keith Wipke, Senior Engineer

CONTACT

Tony Spagnuolo is the Vice President of North American

Operations at Mechanical Dynamics Inc. (MDI). During
his ten years at MDI, Tony has participated in virtual
prototyping deployments at major auto manufacturers
both in Japan and the United States. Before joining
MDI, Tony was employed as an engineer at Rockwell
Automotive Suspension Systems Company (now a
division of Meritor) and General Motors, where he
focused on chassis design and analysis.

Tony can be reach at the following address:

Mechanical Dynamics Inc.

2301 Commonwealthe Blvd.
Ann Arbor, MI 48015
1-734-994-3800

REFERENCES

[1] Bidrawn, Les and Kohnle, Thomas, "The New 3," BMW
Magazine, Vol. 2, 1998, pp. 8-14

[2] Villec, George, Ford Motor Company, “Co-Simulation of an

Automotive Control System Using ADAMS and Xmath”, 1998
International ADAMS Users Conference
 A Logic-Based, Performance-Driven
Electric Vehicle Software Design Tool
David G. Alexander and Donald M. Blackketter, University of Idaho

Co-simulation with ADVISOR and Wave:

Advanced Engine Modeling
Trevor Blohm and Clive Hughes, Ricardo Inc.

Analysis of PNGV Inverter Power and

Current Requirements Using ADVISOR
Gerald W. Davis, Rockwell Automation

Development and Use of a Regenerative

Braking Model in ADVISOR
George Delagrammatikas and Dennis Assanis, University of Michigan

A Systematic Way of Choosing

Driveline Configuration and Sizing Components
in Hybrid Vehicles
Jonas Hellgren, Chalmers University of Technology
 A Logic-Based, Performance-Driven Electric Vehicle
Software Design Tool

David G. Alexander and Donald M. Blackketter

University of Idaho

ABSTRACT specifications based on steady-state operation. It is

written in VisualBasic version 6.0. The governing
The goal of this project was to build a performance- equations are based on the road load power equation.
driven steady-state hybrid electric vehicle design tool Its user-friendly graphical interface provides a platform
using novel equation management and solving routines through which engineers can mix-and-match vehicle
currently under development at the University of Idaho. components and performance goals. The components
The uniqueness of this performance-driven model is in that are modeled include the wheels,
its logic and mathematics based design algorithms. driveshaft/differential, transmission, electric motor,
These algorithms provide advantages when used with battery pack, alternator, and APU. The road load power
traditional numerical techniques. The design algorithms equation includes effects due to aerodynamic drag,
prevent singularity of sets of equations, thereby reducing rolling resistance, uphill climbing and component
the possibility of divergent solutions, while also giving efficiencies.
added flexibility to the user in defining the system of
equations and variables. The algorithms are also used to At the heart of SmartHEV are the logic-based, design
determine the most efficient solution path. This reduces algorithms. These design algorithms have been under
the number of equations that must be solved development at the University of Idaho for the past
simultaneously. Information is also provided to the user several years [1,2]. They have been successfully
to help identify important component relationships. implemented in various software applications including,
Based on the success of this project, these logic and SmartSolve, a linear and non-linear equation solving
numerical techniques could be integrated into software package. The design algorithms are used in
ADVISOR's Autosize feature. The result would be a SmartHEV to determine a valid set of equations and
robust and flexible, performance-driven, hybrid electric unknown variables. Once a non-singular set of equations
vehicle component design and simulation software tool. and their appropriate known and unknown variables
have been selected, the program notifies the user that a
INTRODUCTION solution is possible. The solution is then determined
automatically. Values for all components do not need to
Vehicle simulation software is an essential part of be specified in order to determine partial solutions.
vehicle design and ultimate development. All vehicle
software simulations, however, require that vehicle The algorithms also provide valuable information to the
components be specified prior to executing a simulation. user. Component specifications and performance
This works well if the goal is to gain insight into a parameters are linked to the variables which notify the
particular design. Unfortunately, when the design has user how a change in one variable effects other
not yet been established, running multiple simulations on variables. This way a user can focus in on the correct set
proposed vehicles can be time consuming. ADVISOR of variables in order to modify a design for the desired
has a tool that helps the designer scope out different performance.
vehicle component sizes based on various performance
goals. This feature is known as “Autosize.” While METHODS
providing valuable and accurate design information,
“Autosize” lacks in vehicle details and design flexibility. VEHICLE DESIGN EQUATIONS
SmartHEV was therefore developed to overcome these
limitations. The road load power equation is used to develop the
component relationships. The road load power equation
SmartHEV is a series hybrid electric vehicle software is an accounting of the rate of energy passing through
design tool that calculates vehicle component each vehicle component according to the First Law of
Thermodynamics. The power required at the wheels to DESIGN ALGORITHMS
maintain a vehicle at a prescribed velocity under various
driving conditions is calculated using the following There are three main routines that comprise the design
equation adapted from DOE[3]. algorithms. They are the Variable Select, Solution Path,
and Solution Swap algorithms. The Variable Select
P (t ) = Paero + Prolling + Pgrade + Pacc + Pbearing (1) algorithm is implemented as soon as SmartHEV is
loaded. The Solution Path is used to verify whether the
set of known and unknown variables that was selected is
Where, Paero is the power demand as a result of sufficient for a solution. It also determines the best
aerodynamic drag. Prolling is the load due to the resistance strategy for solving the equation(s). The Solution Swap
of the road on the wheels. Pgrade is the power required to routine is called if a set of equations does not converge
climb a hill. Pacc is the power necessary for acceleration, to a solution, and the unknown guesses and known
and Pbearing is the power required in overcoming the variable values are manipulated in order to approach a
resistance of the bearings and the final drive shaft. valid solution.
The total power, P(t), necessary to meet the velocity is Once SmartHEV has been loaded, the user has the
transmitted to the driveshaft. From this, the torque and option of selecting variables as known, unknown, or
speed requested at the driveshaft can be determined. undecided. Known variables are colored yellow,
unknown variables are colored orange and undecided
Using the torque and speed at the driveshaft, the are white. Undecided variables and the equations in
demand on the transmission is determined. With the which they reside are not used when solving for
transmission gear ratio, the requested torque and speed unknown variables.
from the electric motor are determined. The losses
through the inverter are calculated using a constant The Variable Select algorithm is called whenever a
inverter efficiency coefficient. variable is selected as known or unknown. When a
variable is selected as known, the number of instances
The power requested at the BUS is used to determine that the known variable is present in other equations is
the total power used for discharging and charging of the determined and a decision is made as to whether other
battery pack. The total battery power is, variables are required to be unknown. For example, if
an equation consists of three variables and two of the
Pbat = Palt + Paux + Pbus (2) variables have been specified as known, the remaining
variable would have to be unknown.
Where, Palt is the power from the alternator used to
charge the pack, Pbus is the BUS power demand, and Paux An unknown variable can be changed to known, but only
is the power required for auxiliary loads. Power that is through a process called swapping. During a swap all
discharging from the batteries is positive and negative known variables that interact with the unknown variable
when recharging. Equation 2 balances the power from are presented to the user. The user has the option of
each component. This configuration enables the user to swapping the unknown variable with any on the swap list
determine the power demand from each component or removing, i.e. deselecting, any variable on the list.
while holding the power from the remaining components Swapping simply interchanges the two variables
at zero. The result is the steady state power demand known/unknown status, while removing a known variable
from each component necessary for continuous causes the unknown variable from no longer being
operation. forced as unknown.

The power demand from the road is requested of the After a variable has been selected as known or
battery pack and the alternator. The power through the unknown, the Solution Path algorithm is called. This
alternator is adjusted using a constant alternator routine determines whether the known and unknown
efficiency, and the alternator then requests power from variables constitute a solvable equation set. If it is
the APU. determined that an equation set can be solved, the
Solution Path determines the best solution strategy. This
The battery pack was modeled with an open circuit routine is partially based on work completed by Ramirez
battery voltage, internal resistance and load voltage. A and Vestal [4].
voltage loop equation was written for the circuit. The
power through the circuit was determined by multiply the The solution strategy makes decisions based upon how
voltage loop equation by the current. The current was many unknown variables exist in each equation.
then found using the positive root of the quadratic Equations with only one unknown variable are selected
equation determined from the loop equation with internal as being the first equations to solve. Each equation with
resistance and open circuit voltage held constant. only one unknown is sent individually to the numerical
solving routine. The numerical solving routine consists
of a standard Newton-Raphson implicit method for
systems of equations with a Gauss-Jordan
decomposition routine. The results from the equations
with only one unknown variable are then used in Table 1 lists the results from four different design
subsequent calculations. When systems do not have iterations based on the input parameters and
equations containing only one unknown variable, the performance goals listed in Table 2.
Solution Path routine determines the sequence for
solving that minimizes the number of simultaneous Figure 1. SmartHEV GUI
equations sent to the numerical solver. This provides a
high level of stability and efficiency to the numerical
solving routines.

Finally, the Solution Swap algorithm provides an

additional layer of solving power to SmartHEV. If a set
of equations do not converge to a solution, the Solution
Swap algorithm is called. The Solution Swap algorithm
first determines what the best solution path is for the set
of equations. The best solution path is determined
regardless of which variables have been selected as
known and unknown. The best solution path consists of
a list of known variables that minimizes the number of
simultaneous equations that must be solved at one time.

The user selected known variables are then compared

with the best known variables. If the Solution Swap
routine finds a variable that was selected as unknown by
the user but would be better as a known variable, a
swap is made. The unknown variable then becomes a Table 1. Design results from four different scenarios.
known variable, or swapped-known variable. The
swapped-known variable is assigned a value, typically No Batt. No APU Hybrid I Hybrid II units
one. Since the swapped-known variable was originally
an unknown variable, its value is arbitrarily selected. The Road 33,835 33,835 33,835 33,835 watts
value of the known variable that was swapped, now the Load
swapped-unknown variable is stored. Accel. 3,478 3,478 3,478 3,478 watts
Power
The Solution Swap then attempts to solve the modified Grade 20,463 20,463 20,463 20,463 watts
set of equations. If the modified set of equations Aero. 5,989 5,989 5,989 5,989 watts
converges to a solution, the results are compared. If the Rolling 3,065 3,065 3,065 3,065 watts
solution of the swapped-unknown variable is the same Bearing 838 838 838 838 watts
as its stored value, the routine stops because a solution
Motor 35,616 35,616 35,616 35,616 watts
was found. In most cases, the stored value and the
Power
solved value will be different because the solved value is
Motor 92.7 92.7 92.7 92.7 rev/s
based on the swapped-known variable with a value that
Speed
was arbitrarily selected. The swapped-known variable is
adjusted and the set of equations is solved again. These Motor 61.2 61.2 61.2 61.2 Nm
two adjustments and resulting solutions are then used to Torque
perform a linear interpolation in order to get a better APU 51,400 0.0 34,557 30,000 watts
value for the swapped-known variable. Adjustments are Power
made until the swapped-unknown variable value agrees APU 110 0.0 100 80 rev/s
with the stored value within a specified tolerance. Speed
APU 73.9 0.0 55 60 Nm
RESULTS Torque
APU Eff. 0.25 0.0 0.19 0.20
Figure 1 shows the GUI of SmartHEV. The variable Batt. 0.00 48,830 16,000 20,330 watts
value boxes have adjustments that change the value by Power
10% for every click of the up or down arrow. Once the Batt. 0.0 245 80 102 amps
variable has been changed, the background color Current
becomes yellow. If a variable is forced to be unknown its Alt. Power 48,830 0.0 32,829 28,500 watts
background color turns orange. The variable value with Alt. 244 0.0 164 143 amps
the dark background and light numbers in Fig. 1 indicate Current
that the variable has changed value because of a Gear ratio 6.67 6.67 6.67 6.67
change that was made to a known variable value. This is
particularly helpful when changing known values In Table 1, the first column is the results with no power
because it indicates which variables are effected by the provided by the battery pack. Column two lists the
change.
results with no power provided by the APU. The third Incorporating SmartHEV into ADVISOR would result in a
column, Hybrid I, lists the model results based on a powerful vehicle design and simulation environment.
hybrid operation. Neither the APU nor the Battery pack SmartHEV could be used to investigate preliminary
output power was specified. Hybrid II, the fourth column, vehicle designs. The flexibility of SmartHEV and its user-
was designed with a known APU power output of 30 kW. friendly GUI provides tremendous insight into component
As a result, the APU speed and were adjusted to parameter interactions and relationships. A preliminary
accommodate the constant APU power output. design can be easily modified and adapted in order to
determine the best configuration. The best vehicle
Table 2. Design Parameters design could then be imported into ADVISOR and used
to run extensive simulations. SmartHEV could also be
Mass 1413 kg used to redesign components since it does not require a
2
Accel. 0.1 m/s complete design to be established before determining
Velocity 24.6 m/s results. Individual components can be modified and
Grade 0.06 tested in SmartHEV and then incorporated into an
Drag Coef. 0.335 overall vehicle simulation in ADVISOR.
2
Frontal Area 2 m
SmartHEV is a performance-driven hybrid electric
Rolling Coeff. 0.006 vehicle model that provides flexibility and power to the
Bearing Loss 9.6 Nm design engineer. The unique logic-based algorithms
Wheel Radius 0.282 m allow many different combinations of performance
Batt. Voltage 200 volts variables and constraint variables to be selected with the
Batt. Resistance 0.0045 ohms added security of maintaining a solvable set of
equations. If a particular arrangement of known and
The next scenario sizes the vehicle components based unknown variables is difficult to solve, SmartHEV swaps
upon the following performance goals, constant grade of variables in order to find a solution path that is solvable.
6%, maximum speed of 40.2 m/s, and acceleration of 0- Information provided to the user while selecting variables
26.9 m/s in 12 sec. The Because SmartHEV is a and component parameters is extremely helpful and
preliminary vehicle design tool and not a vehicle insightful. Knowing which component is effected by a
simulation tool, the performance criteria had to be change in a particular variable guides the user towards a
modified slightly. SmartHEV determined the vehicle workable design solution. The results are fast, accurate,
component sizes for each performance criteria and easy to follow. Incorporating SmartHEV technology
separately. The grade test was performed under into ADVISOR would result in an invaluable hybrid
constant speed at 24.6 m/s (55 mph). SmartHEV then electric vehicle design and simulation tool.
calculated the component sizes based on the maximum
speed at zero grade. The best effort acceleration was CONTACT
determined at zero grade as well. The results were
compared and the maximum component size was David G. Alexander is a graduate student studying
determined. The same design parameters as Table 2 Mechanical Engineering at the University of Idaho. He
were input to both SmartHEV and ADVISOR. These completed a Bachelor of Science in Physical Science
results are compared with the results of a similar from California State University, Chico in 1992, and a
simulation run in ADVISOR, see Table 3. Master of Science in Mechanical Engineering from the
University of Idaho in 1999. He is currently focusing his
Table 3. Component Power Requirements for efforts on developing a hybrid electric vehicle energy
SmartHEV and ADVISOR management control strategy for the University of
Idaho’s FutureTruck 2000 Suburban. Mr. Alexander’s
Component SmartHEV ADVISOR email address is [email protected].
APU 46 kW 49 kW
Battery Modules 29 @ 12 V 29 @ 12 V Dr. Donald M. Blackketter is an associate professor at
the University of Idaho. His expertise is in equation
Alternator 44 kW 56 kW
management techniques for systems of linear and
nonlinear sets of equations. He has extensive
CONCLUSION experience in dynamic and kinematic modeling of
machine components. Dr. Blackketter’s email address is
SmartHEV calculates vehicle component sizes similar to [email protected].
those of ADVISOR. However, the two programs have
fundamental differences. ADVISOR is an excellent
vehicle simulation tool, whereas, SmartHEV excels in
vehicle and component design. Since SmartHEV is
based on steady state performance its inputs were
modified in order to compare it with ADVISOR.
REFERENCES

1. Glumbik, J. P., “Variable Interactions within Design

Equations: A Methodology in Equation Management,” MS
Thesis, University of Idaho, 1997.
2. Benson, J. L., “Simultaneous Equation Algorithms for
Variable Interaction in the Solution Path,” MS Thesis,
University of Idaho, 1998.
3. DOE, “SIMPLEV-A Simple Electric Vehicle Simulation
Program,” Version 2.0, DOE, Idaho Falls, ID.
4. Ramirez, F. W., “Algorithms for Structuring Design
Calculations,” Chemical Engineering Science, 1972, Vol.
27, pp. 2243-22.
 Co-simulation with ADVISOR and Wave: Advanced Engine Modeling
by

Trevor Blohm, Clive Hughes

Ricardo, Inc.

Ricardo is currently creating a link between ADVISOR and WAVE, Ricardo's industry-
standard engine performance and 1-D gas dynamics software package. WAVE models
are based on detailed engine dimensions. The resulting co-simulation will allow
prediction of drive cycle performance for engines without performance data.
Furthermore, the effect of modifications such as turbocharger size and valve timing can
be examined.

Ricardo will present the details of the ADVISOR-WAVE link, including the required
modifications to ADVISOR. Fuel consumption of a Class 7/8 truck modeled with and
without this link will also be presented.
 Analysis of PNGV Inverter Power and Current Requirements
Using ADVISOR
Gerald W. Davis
Advanced Technology Labs, Milwaukee, Rockwell Automation

ABSTRACT through a one-step recursion estimates the actual

performance in response to a requested performance. If
The PNGV Program is composed of a number of the actual performance and the requested performance
research projects. The inverter section of the PNGV are identical, then the vehicle was able to satisfy the
program is called the “Automotive Integrated Power input “Drive Cycle”.
Module” (AIPM). As part of the AIPM research activity,
ADVISOR was used to better understand the power and To use ADVISOR, the user interacts with three screens.
current requirements of the PNGV vehicle operating The first screen inputs a vehicle configuration. Fig. 1
under various drive cycles. Since the AIPM shows an example vehicle input screen. The second
manufacturing cost is a key variable for the research screen lets the user specify the Simulation Parameters
project ADVISOR was used to determine an appropriate which determine what “drive cycle” and grade. The third
battery bus voltage, and peak current limits. screen is the Results screen which displays how the
vehicle performed.
INTRODUCTION

AIPM specifications [1, 2, 3] require a peak power of 55

kW for up to 10 seconds. This requirement has a direct
impact on the size of the power silicon devices, and the
thermal requirements of the AIPM. If the power
requirement can be reduced while maintaining the
vehicle performance requirements, the AIPM will have
reduced size, thermal requirements, and cost.
Therefore, it was felt that the AIPM power requirement
had to be investigated in detail. Along with the power
requirement, the AIPM battery bus voltage and motor
current requirements were also investigated.

The objective was to see if it would be possible to refine

the requirements for Peak Power, and the Full
Performance Voltage Input Range to values that reduce
the size and cost of the AIPM. ADVISOR was selected Fig. 1. ADVISOR: Vehicle Input screen of ADVISOR
to perform the vehicle configuration and energy balance
simulation. First, consider the Vehicle Input screen shown in Fig. 1.
Each vehicle component, e.g. Fuel Converter (engine,
THE PNGV VEHICLE fuel cell, etc.), Motor Controller, Generator, etc., has
information about the component’s thermal
ADVISOR is an energy balance and efficiency simulation characteristics, energy conversion efficiency, and weight.
developed at the National Renewable Energy Laboratory To understand the components and the underlying
(NREL) in Golden, CO [4]. The simulation computes the variables of ADVISOR, the best approach is to read
energy required to perform a vehicle “Drive Cycle” task reference [4], and to peruse ADVISOR’s HELP files. The
for a given time duration and flows the energy demand vehicle components shown in Fig. 1 are estimates of the
through the vehicle sub-components that must satisfy the component characteristics for a PNGV vehicle. The
demand. If the requested energy demand exceeds the VEH_PNGV.m file contains these values and was
sub-components capability to supply it, the simulation supplied by NREL. The MC_55AIPM was a first cut
has a backward path that communicates what is possible estimate of the AIPM inverter/motor system and was
to the previous sub-component. In this way, ADVISOR derived from the MC_PM58.m file.
A key variable to note in Fig. 1 is the total vehicle weight. The results screen shows the performance of the user
During an analysis, it is common to adjust the sizes of specified vehicle to the user specified drive cycle. Fig. 3
various components, e.g. the inverter or engine sizes. A shows an example screen. A large selection of output
change in power size also impacts the weight of the variables can be displayed in any of the four plot
component, so care must be taken to make sure that the windows on the left. In Fig. 3, the first plot shows the
vehicle weight doesn’t change between simulations. requested speed of the drive cycle was matched by the
response of the vehicle, i.e. the energy delivery
SELECTION OF PERFORMANCE CRITERION performance was adequate for the drive cycle. The
second plot in Fig. 3 shows the State of Charge (SOC) of
The second screen in an ADVISOR simulation is the the battery. In this example, the battery never
Simulation Parameters screen. This screen specifies the discharged to a point where the Internal Combustion
performance test that will be applied to the vehicle Engine (ICE) was required. The third plot shows the
determined in the Vehicle Input screen. An example is RMS current delivered by the battery. The fourth plot
shown in Fig. 2. shows the RMS Power Bus voltage.

It must be remembered, that ADVISOR only shows RMS,

DC energy values for the Motor Controller (MC) part of
the simulation. Hence, the VAR current component of a
three-phase system is not shown. This means that the
actual currents flowing in a three-phase system would
have to be estimated by assuming a power factor to the
motor.

Fig. 2. Simulation Parameters screen of ADVISOR

The green section in the upper right of the screen shown

in Fig. 2 selects a drive cycle. A drive cycle is a set of
speed commands arrayed over a sampled set of one
second time intervals. The example in Fig. 2 shows the
FUDS (Federal Urban Driving Schedule) drive cycle.
Fig. 3. Results screen from ADVISOR.
A road grade of 6.5% has been selected ([email protected]%,
as specified in [1]) in the blue, middle section of Fig. 2. The example in Fig. 3 shows that the FUDS drive cycle
The Parametric Study section in the lower right of the was satisfied with RMS currents which didn’t exceed 200
screen is not checked, so this option doesn’t occur in the A. The peak currents occur during periods of maximum
simulation. The display window on the left half of the acceleration when the power factor is close to a value of
screen allows the user to view various parameter arrays 1.0. Hence, a peak current estimate of 300 A provides
selected by the user. reasonable margin for demand.

After completion of the Simulation Parameters screen, The power requirement is the fourth plot window in Fig. 3
the user clicks the RUN button at the bottom right. This and can also be viewed by plotting the “ess_power” array
action causes ADVISOR’s Results screen to appear. as shown in

RESULTS FOR VARIOUS SIMULATION SCENARIOS Fig. 4. The plot shows that the peak value required by
the FUDS drive cycle was close to 55 kW, but expansion
This section looks at the simulation results for testing the of the plot shows that the peak value represented only a
PNGV vehicle with the FUDS and US06 drive cycles. single point
The FUDS drive cycle is the likely certification drive cycle
for HEV vehicles. The US06 drive cycle is a higher road A more informative view of the power requirements is
speed drive cycle and may be applied to HEVs at some seen in Fig. 5. This is a histogram of the power samples
point in the future. delivered by the motor controller system to satisfy the
FUDS drive cycle. The bin width in the histogram is Fig. 7 shows a similar power occurrence histogram, but
about 400 W, and the large peak centered at 0 W is due as applied to the US06 drive cycle with 6.5% grade data.
to the zero speed command sections in the drive cycle. The histogram shows a high density of power level
Note that the vehicle accessories file was adjusted for occurrences around 60 kW. The difference between the
this particular simulation to draw zero power. A plot of requested speed command and actual speed command
the histogram of power values without the large peak is was also significant as shown in Fig. 8. Hence, the 55
shown in Fig. 6. kW inverter would not satisfy the US06 drive cycle with
the current PNGV vehicle weight.
RMS Power, Watts

Counts
Seconds
Power, Watts
Fig. 4. RMS Power requirement for results shown in Fig.
3. Fig. 7. Histogram of Motor Controller power occurrence
density vs power level derived from the US06 at 6.5%
performance data.
Counts

Power, Watts

Fig. 5. Histogram of Motor Controller power occurrence

density vs power level derived from the FUDS at 6.5%
performance data.

Fig. 8. Difference between Requested Speed and Actual

Speed for US06 drive cycle at 6.5% grade.
Counts

VARIATION OF BATTERY BUS VOLTAGE

As a battery discharges, the voltage decreases. The

battery used for most of the ADVISOR simulations was a
Power, Watts 26 cell NIMH battery that produced a nominal voltage of
Fig. 6. Histogram of Motor Controller power occurrence 364 volts. This battery configuration was selected to
density vs power level for densities above the peak shown produce a power bus voltage similar in magnitude to that
in Fig. 5. believed to be used in the Chrysler and General Motors
HEV prototypes.
Fig. 6 shows that the peak values occur at about 55 kW
(actual max value is 54.06 kW). Hence, a 55 kW power Fig. 9 shows the results of a [email protected]% drive cycle
delivery of power from the battery and motor controller run with the battery initiated to a low State of Charge
subsystems of the PNGV vehicle configuration satisfies (SOC). Even under these conditions, the simulation
the FUDS drive cycle with a 6.5% grade. indicates that the drive cycle is satisfied right up to SOC
= 0.0 (around t = 740 sec) at which point the simulation
ends. Is this an artifact of ADVISOR? Apparently not. REFERENCES
NIMH maintain reasonable voltage even at very low
SOC. In Fig. 9, we see that the pb_voltage is dropping 1. US DoE Solicitation No. DE-SC02-98EE50525,
as the SOC decreases. This amount of voltage drop has “Research, Development, and Demonstration for an
Automotive Integrated Power Module (AIPM) for a New
been determined from the manufacturer’s data sheets Generation of Vehicles”, August 19, 1998.
(Ovonic, Inc.). 2. “POWER ELECTRONICS MODULE INTERFACE
STANDARD PURPOSE OF THIS ACTIVITY”,
Presentation 2/24/97, ATLANTA, GA by Sunil Chhaya
GM-Advanced Technology Vehicles.
http://grouper.ieee.org/groups/1461/Comments_and_Stra
wman_SPEC.htmlonal section.
3. “Automotive Integrated Power Module” Allan R. Gale
(Ford Motor Co.), Sunil Chhaya (General Motors, Cyrus
Ashtiani (Chrysler), David Hamilton (US DoE),IEEE P1461
Workgroup Mtg, IEEE IAS 97, New Orleans, 10/7/97.
4. “ADVISOR 2.1: A User-Friendly Advanced Powertrain
Simulation Using a Combined Backward/Forward
Approach”, K. B. Wipke, M. R. Cuddy, and S. D. Burch,
NREL/JA-540-26839, Sept., 1999.
http://www.ctts.nrel.gov/analysis/reading_room.html

Fig. 9. Battery discharge effect on the power bus.

CONCLUSIONS

The 55 kW design satisfies the FUDS drive cycle almost

exactly, but does not satisfy the US06 drive cycle. EPA
certification requires that vehicles satisfy the drive cycle
speed command within 5%. ADVISOR simulations
indicate that, indeed, a 55 kW inverter is about the
correct size for the PNGV vehicle weight to satisfy the
[email protected]% drive cycle.

ACKNOWLEDGMENTS

I would like to thank Jim Merrit, Keith Wipke, and Valerie

Johnson for their assistance in getting and using
ADVISOR.

CONTACT

Gerald Davis, Power and Drives Lab, Advanced

Technology Labs, Rockwell Automation, Milwaukee,
WI 53204. Phone: (414) 382-4125. e-mail:
[email protected].
 Development and Use of a Regerative Braking Model in ADVISOR
by

George Delagrammatikas, Dennis Assanis

University of Michigan

The poster details our development of a modular, downloadable regenerative braking

control strategy for a parallel HEV. The main objective behind this work was to be able
to predict how our new strategy would perform in ADVISOR, then download it to the
vehicle controller for on-road use and validation. Comparisons between the two models
were then made and conclusions were drawn regarding their differences in various
driving cycles and component configurations.
A SYSTEMATIC WAY OF CHOOSING DRIVELINE
CONFIGURATION AND SIZING COMPONENTS IN
HYBRID VEHICLES
JONAS HELLGREN
MACHINE AND VEHICLE DESIGN
CHALMERS UNIVERSITY OF TECHNOLOGY
GÖTEBORG, SWEDEN
 AIM WITH THE METHOD
Transform demands and conditions to suitable drivelines (con-
ventional/hybrid)
Definitions:
• Driveline=vehicle configuration with sized components
• A driveline may either pass or not pass demands. Examples of
demands are driving cycle and use of emission free zones.
• Conditions makes a driveline more or less suitable and varyies
with time and depends on country. Some conditions are fuel price,
tax on pollution and discount rate.
• Name of method = Driveline Synthesis (DS).
• DS is of approximative nature, i.e. the result is not a definitive
answer of how to design a vehicle, but more an advise.

CHALMERS UNIVERSITY OF TECHNOLOGY

 DESCRIPTION OF METHOD FOR CHOOSING
OF PROPULSION SYSTEM
method

Individual
Conditions Optimization Evaluation
Suggested
Cost of individual(s)
Demands individual

Library with available Cost

technology weights
• Many combinations => a simple and fast evaluation is needed.
• Suitable individuals with defined driveline configuration and con-
trol strategy are finally presented.

CHALMERS UNIVERSITY OF TECHNOLOGY

 OPTIMIZATION METHOD
The reasons for choosing genetic algorithms are:
• A non-gradient based method is needed, because there is a non
continuos relation between cost and vehicle properties.
• If some type of driveline configuration is superior this will proba-
bly be known early.
• Favorite (good) drivelines can be stored and reused, i.e. compete
under different circumstances.
• Genetic algorithms suit very well with parallel (super) computers,
which may be needed if more drivelines will be covered.
On a big scale, genetic algorithms work by the principle of evolution:
the best-adapted individuals survive and reproduce.

CHALMERS UNIVERSITY OF TECHNOLOGY

 fitness=1/
=individual
Cost

Indiv.nr.

• Several individuals (search agents) evaluate the search space in

parallel. The individuals mutate and mate with each other.
• Described by the genes of the individual: type of configuration,
mass of PPU, mass of storage, fluctuation of State Of Charge
(SOC) and type of strategy.

CHALMERS UNIVERSITY OF TECHNOLOGY

 EVALUATION
To evaluate an individual, the following goal function is considered:
cost 1
[Euro/km] = ----------------
- =
fitness
( w c ⋅ comp + w f ⋅ fuel + w g ⋅ grid + w p ⋅ pollution ) ⁄ performance

TABLE 1.
comp Cost of components.
fuel Fuel cost.
pollution Cost of pollution.
grid Cost of grid energy.
wc, wf, wg, wp Cost weights.
performance How well the driving cycle is managed. 1 if the vehicle can follow
the desired speed else 0.1 or 0.00017 etc. .

CHALMERS UNIVERSITY OF TECHNOLOGY

LOAD CASE SIMULATION
• The principle is that one follows the system in the time domain.
The driving cycle is divided into load cases. A load case can, e.g.
be an acceleration from 0 to 80 km/h in 15 s.
• Load case simulation gives much faster computation, approxi-
mately 1000 times faster compared to a conventional simula-
tion.
• For each load case following can be determined from a vehicle
model: fuel consumption, emissions, wear of storage and change
of state of charge.
• Price of pollution:
price
of pollution

grams of
pollution per
km
limit

CHALMERS UNIVERSITY OF TECHNOLOGY

CASE STUDY FOR CHOOSING OF PROPULSION SYSTEM
City bus and Low speed driving cycle with many starts and stops
(R = ranking).
R Driveline configuration mass PPU mass Stor. cost
[kg] [kg] [Euro/km]

1 Series HEV: 200 1250 0.67

diesel engine+super-capacitor (100 kW) (1250 kW)
2 Conventional: 400 - 0.73
diesel eng./automatic tr. (200 kW)
3 Series HEV: 400 500 0.82
diesel engine+NiMH battery (200 kW) (500 kW)
4 Series HEV: 100 1250 0.89
fuel cell+super capacitor (50 kW) (1250 kW)

CHALMERS UNIVERSITY OF TECHNOLOGY

PART COSTS/EVALUATION OF DRIVELINES

Series HEV => lower

fuel consumption and
higher capital cost.
Conventional vehicle
=> higher fuel consump-
tion and lower capital
cost.

CHALMERS UNIVERSITY OF TECHNOLOGY

 CONCLUSIONS/DISCUSSION
• During the application of DS, it has been found that the most
appropriate driveline is very dependent on the demands and condi-
tions which are made.
• It is pointless to say that one driveline configuration, in general, is
better than another. Eg. very high cost on pollution or a cheap fuel
cell will make fuel cell vehicles profitable.
• It is not enough to only consider maximum power request. Case
1=driving cycle contains extremely many starts and stops => super
capacitor is profitable . Case 2=fewer starts and stops => NiMH
battery is profitable. This is due to less wear in storage.
• A preliminary result is that a cheap battery or super capacitor will
make HEV very competitive.

CHALMERS UNIVERSITY OF TECHNOLOGY

 FUTURE WORK
This should be done in the future to improve and evaluate DS further:
• Further validation of DS. ADVISOR is a candidate of doing this.
• Make DS more accurate.
• Add components and driveline configurations. Data from ADVI-
SOR is interesting.
• Collect accurate data on different components used in driveline
configurations. Data from ADVISOR is interesting.
• Improve control strategy for each driveline, an adaptive method
(learn while driving) is one way.

CHALMERS UNIVERSITY OF TECHNOLOGY

 ASSIGNER/LANGUAGE/PLATFORM
• The work is assigned of Volvo Buses and is specially focusing on
heavy vehicles.
• The tool is written in Matlab, by using Matlab:s own program-
ming language.
• At present the tool is not available for the public or for sale.

CHALMERS UNIVERSITY OF TECHNOLOGY