Trade‐Offs Between Fuel Economy and NOxTrade Offs Between Fuel Economy and NOxEmissions Using Fuzzy Logic Control With a 

Hybrid CVT ConfigurationHybrid CVT Configuration

Aymeric Rousseau, Sylvain Saglini, Mike Jakov, Donald Gray and Keith Hardy

Center For Transportation Research

Sponsored by B. KostU.S. Department of Energy

This presentation does not contain any proprietary, confidential, or otherwise restricted information

Tools for Integrated Development

Simulation Validation

Emulation

I/O Board

Motor

Measurements

CommandsI/O Board

Motor

Measurements

Commands

Component/subsystem models used in vehicle simulation must be li bl t l t d d l hi l i t

Brake

Dynamometer

CVT

oto

EngineBrake

Dynamometer

CVT

oto

Engine

applicable to emulated and real vehicle environments– Forward models for realistic behavior– Comparable inputs/outputs to hardware

Validation is critical– Validation is critical

PSAT Is Flexible & Reusable

Drivetrains constructed from user choices

Numerous configurations can be explored(>130: conventional, electric, fuel cell, parallel, series, power split...)

User friendly graphical user interfaceUser-friendly graphical user interface

Easy integration of new component data, models and control

Model format is generic (3 inputs / 3 outputs)Model format is generic (3 inputs / 3 outputs)

PSAT Looks ForwardForward modeling (driver-to-wheels) more realistically predicts system dynamics, transient component behavior and vehicle responseresponse.Commands from a Powertrain Controller to obtain the desired vehicle speedCommands from a Powertrain Controller to obtain the desired vehicle speed

Engine Cl t h T i i Fi l D i Wh l V hi l

More accurately represents component dynamics (e.g. engine starting and warm-up, shifting, clutch engagement ...)Allows for advanced (e g physiological) component models

Engine Clutch Transmission Final Drive Wheel Vehicle

Allows for advanced (e.g. physiological) component modelsAllows for the development of control strategies that can be utilized in hardware-in-the-loop or vehicle testingS ll ti t hSmall time steps enhance accuracy

Vehicle Characteristics Summaryy

Pre-transmission parallel HEV:

Vehicle Mass 1297kgFA 1.5 m2Cd 0.2Engine 1.7L MB CIDIMotor 32kW PM UQMCVT JATCO CK2-CVT Battery Li ion 14Ah 96 cellsBattery Li-ion 14Ah 96 cells

Trade-off Between Fuel Economy and Emissions

APRF A Class 1.7L DataUQM Data

NN ModelModels

1

1.2

DIRECT Optimization

APRF0 01 02 03 04 05 06 07 08 09 1

0

0.2

0.4

0.6

0.8

1

Deg

ree

of M

embe

rshi

p Optimization

HILCVT Data

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1Driver Input

Fuzzy Logic Control

State-of-the-art Test Facility Used For Engine TestingEngine Testing

Dynamometer capability: Highl transientHighly transient220 kW, 1067 Nm, 8000 rpmInertia 1.5 kg-m^2

Emission Equipments. Pierburg AMA-2000 raw bench Cambustion fast HC and NO

– Two channels per species– Less than 5 msec response

timetime TEOM (Tapered Element Oscillating

Microbalance)

NN NOx Emission Demonstrates lBetter Correlation

Test data

Neural Network

Steady state

Neural Network

Fuzzy Logic Controller Overview

Driver Input [ 1 to 1]

Engine Command [0 to 1]

Fuzzy Logic

[-1 to 1]

Battery SOC [0 to 1] Motor Command [-1 to 1]

Fuzzy Logic Control Strategy

Motor SpeedEngine Speed

StrategyCVT Gear Ratio

Friction Braking [0 to 1]

Vehicle Speed

Membership Function For Battery f hState of Charge

1 2 Too Low Low Normal Too High

0 8

1

1.2

rshi

p

Too Low Low Normal Too High

0 4

0.6

0.8

e of

Mem

ber

0

0.2

0.4

Deg

ree

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

0

State of Charge

Strategy Defined By 27 Rules

If driver input demand is medium andpSOC is normal andMotor speed is optimal andCVT ti i t th l t d

Control inputs are used to define the condition

CVT gear ratio is not the largest andvehicle speed is not lowThen Engine speed is 375 rad/s andThen Engine speed is 375 rad/s andEngine command is 0.76* andMotor command is –0.25

Outputs are defined

* Means we ask for 76% of the best efficiency curve

Fuel Economy vs. NOx Emission C t l St tControl Strategy

Best Efficiency Curve

Area of OptimizationArea of Optimization

Best NOx Curve

FE and NOx Can Vary From +/- 10% B d U C t l St tBased Upon Control Strategy

Trade off between Fuel Economy and NOx Emissions

60

70

80

40

50

60

10

20

30

0FUDS Hybrid eq (mpg) NOx - FUDS (g/miles*100) FHDS hybrid eq (mpg) NOx - FHDS(g/miles*100)

Rule Based (SOCgoal = 0.5) Rule Based (SOCgoal = 0.7)Fuzzy Fuel Economy (SOCgoal = 0.7) Fuzzy Emission (SOCgoal = 0.7)

Hybridization Increases FE By Using The Best Engine Efficiency CurveBest Engine Efficiency Curve

Maximum Torqueq

Conventional Vehicle Operating Points

Hybrid Electric Vehicle Operating Points

Without Basic Changes In The Powertrain, Hybridization Improves FE, but May Increase NOxHybridization Improves FE, but May Increase NOxEmissions !

HEVVehicle With No Basic Changes

Optimized System

ConventionalVehicle

System

Perspectives

Fuel economy and emission trade-off methodology has been demonstratedbeen demonstrated

System optimization is needed to resolve the diesel emission issues (because engine control might not be ( g gsufficient).

Preliminary simulation results are dependent on components and driving cycle and need to be validated using HIL