significant reduction of powertrain calibration effort in production programs avl en

TM Symposium ChinaApril 24 – 26, 2012, Beijing, China

Gerhard Kokalj, AVL List GmbH, Graz 1

Significant Reduction of Powertrain Calibration Effort in Production Programs

Automated calibration of driveability on chassis dyno and powertrain testbed

IT’S TIME TO CHANGE

breakthrough in calibration technology

Gerhard Kokalj,

Head of Transmission, Hybrid and e-Vehicle Calibration



Typical Customer Quotes:

1. We are not able to cover all

potential calibration variants within

the available time frame…

2. We have to deliver variant

calibrations with fewer and fewer

prototype vehicles…

3. Our calibration methodology needs

to change to address these critical

issues…

Time saving of 25%

per vehicle variant

Reduction in vehicle usage

of 30% per vehicle variant

ACT™ represents a new

automated calibration

methodology



We automated the process

We employed a global model-based

approach with objective driveability rating

We developed calibration tools,

methodologies and processes



Calibration ProcessTest Plan Phase

Test Plan

DoE: same information, fewer measurements, driveable range calculation

Desktop

Test Plan

calibration phase (office)

testing phase (vehicle)



Test Plan PhaseObjective: e.g. Gearshift Quality Calibration

Example showing the number of parameters to be

optimized for a single gearshift

Time

Clu

tch

sw

itch

off

Tu

rbin

e

Sp

eed

Clu

tch

sw

itch

on

To

rqu

e

Dem

an

d

Gear

Sh

ift

Calibration Task:Find the best combination of parameters P1, …, Pn

Time

Clu

tch

sw

itch

off

Tu

rbin

e

Sp

eed

Clu

tch

sw

itch

on

To

rqu

e

Dem

an

d

Gear

Sh

ift

TimeTimeTime

Clu

tch

sw

itch

off

Tu

rbin

e

Sp

eed

Clu

tch

sw

itch

on

Clu

tch

sw

itch

on

To

rqu

e

Dem

an

d

To

rqu

e

Dem

an

d

To

rqu

e

Dem

an

d

Gear

Sh

ift

Gear

Sh

ift

Calibration Task:Find the best combination of parameters P1, …, Pn

Increasing number of parameters

Use of hybrid and engine signals in a complete system concept

Why is it so complicated?



Test Plan PhaseGeneration of Test Plan with DoE

IV. Calculation of Test Plan using AVL CAMEO

I. Definition of speed (engine / turbine) and pedal steps / turbine torque as variation parameters (global test plan)

II. Definition of TCU / ECU variation parameters

– parameters for Shift Quality (e.g. clutch pressure, torque reduction)

– parameters for Tip In / Tip Out (e.g. open loop air control, time constant filter)

III. Definition of Constraints



Test Plan


Desktop

Fully automated night and weekend test runs on vehicle chassis dyno or powertrain testbed Vehicle

Test Plan

Test Run

Calibration ProcessTest Run Phase





Test Run Phase (Vehicle Chassis Dyno)Acceleration signal derived from a force sensor

Fixing Frame and longitudinal force

sensor

AVL-DRIVE™ for drivability quality

evaluation

AVL CAMEO Transmission™ is used

• to create test plans using DoE methods

• to perform controlled automated test runs

• for data evaluation and optimization

according the specified drivability target

AVL DriCon™ enables fully automated

and over night test



Driveability Simulation model

AVL-DRIVE™ for drivability quality

evaluation

AVL CAMEO Transmission™ is used

• to create test plans using DoE methods

• to perform controlled automated test runs

• for data evaluation and optimization

according the specified drivability target

AVL DriCon™ enables fully automated

and over night test

Test Run Phase (Powertrain Testbed)Acceleration signal derived from a driveability model



Test Run PhaseSteps of the Automated Test Run

Speed

To

rqu

e P1

P4

P3

P2

TCU Parameters

Starting from the generated test plan,

CAMEO Transmission™ defines the test parameters

1

AVL-Drive™ analyzes each gearshift event

2

AVL-DRIVE™ test results for each test variation

TurbineSpeed TurbineTorque ShiftType P1 P2 P3 P4 DR-Gearshift DR-GearshiftCOMFORT DR-GearshiftDYNAMIC DR-Jerks DR-Kick DR-Shiftdelay DR-Shiftduration flareduration shift_delay shift_duration

[rpm] [Nm] [-] [Nm] [bar] [bar] [ms] [-] [-] [-] [-] [-] [-] [-] [-] [-] [-]

Variation Variation Response Variation Variation Variation Variation Response Response Response Response Response Response Response Response Response Response

3194 153 23 -84 1,22 1,02 370 7,26 6,93 9,06 9,94 9,98 7,96 9,28 0,79 0,55 0,33

3194 153 23 -84 2,30 1,02 570 6,98 6,47 8,63 8,98 9,18 7,78 9,24 0,92 0,63 0,3

3194 153 23 -84 2,57 1,12 570 6,92 6,38 7,8 9,94 9,98 7,79 9,2 0,9 0,63 0,31

3194 153 23 -84 2,57 1,42 260 6,23 6,01 4,98 8,97 9,17 7,77 9,35 0,87 0,64 0,24

3194 153 23 -84 2,84 1,63 260 5,8 5 5,58 9,94 9,98 7,82 9,48 0,84 0,61 0,2

3194 153 23 -72 2,30 1,22 570 6,07 5,71 4,75 7,54 8 7,79 9,12 0,85 0,63 0,3

3194 153 23 -72 2,84 1,02 260 7 6,46 7,85 8,72 8,95 7,74 9,19 0,95 0,65 0,31

3194 153 23 -66 1,22 1,42 570 6,32 5,45 7,45 9,94 9,98 8,11 9,29 0,72 0,49 0,26

3194 153 23 -54 2,03 1,32 420 6,25 5,85 5,68 9,94 9,98 7,89 9,15 0,81 0,58 0,29

3194 153 23 -48 1,22 1,32 260 6,41 5,45 8,46 9,94 9,98 7,96 9,28 0,81 0,55 0,27

3194 153 23 -36 1,22 1,02 420 6,67 6,4 6,09 9,94 9,98 8,11 8,65 0,72 0,49 0,47

3194 153 23 -36 2,57 1,63 420 5,55 4,88 4,88 9,94 9,98 7,83 9,34 0,81 0,61 0,23

3194 153 23 -30 1,22 1,02 570 7,06 6,59 8,96 9,94 9,98 7,94 8,91 0,83 0,56 0,42

3194 153 23 -30 1,22 1,32 520 6,34 5,56 6,2 9,94 9,98 8,13 9,04 0,71 0,48 0,32

3194 153 23 -30 1,49 1,02 260 6,07 5,19 6,13 8,99 9,19 7,94 8,86 0,78 0,56 0,39

3194 153 23 -30 2,30 1,63 520 6 5,14 6,71 9 9,2 7,91 9,36 0,75 0,57 0,23

Variation Parameters Measured results from AVL-DRIVE™

Results (test result list) are stored in CAMEO Transmission3

Parameter Settings and AVL-DRIVE™ test results for each test are stored in one line

Start of Post Processing:

After plausibility check these results are

used for response model generation

Parameter 1

DR

-Ge

ars

hif

t R

ati

ng

DR

-Sh

iftd

ura

tio

n R

ati

ng

4



Test Plan


Desktop


Statistical data analysis

Desktop

Test Plan

Test Run

Data

Check

Modeling &

Optimization

Map

Generation



Calibration ProcessStatistical Analysis Phase



Statistical Analysis PhaseOutlier Suppression



Test Plan


Desktop



Desktop

Test Plan

Test Run

Data

Check

Modeling &

Optimization

Map

Generation

Optimizes Control Unit Parameters,Global modeling allows multiple calibration to be produced offline (e.g. sport, comfort) Desktop



Calibration ProcessModeling and Optimization Phase



Parameter Value

AVL-DRIVE

“Gearshift

Rating”

Modeling and Optimization Phase Why do we need models?

Mathematical Model

Models describe the influence of each parameter value

on each response value



Modeling and Optimization Phase Result of Model Generation

The results (e.g. shift

quality rating) for each test

variation (e.g. gearshift)

are fitted with a

mathematical model to

describe the influence of

each parameter

on each response.

DR

-Ge

ars

hift R

atin

gD

R-S

hiftd

ura

tion R

ating

Model 2„shiftduration“

Overview of parameters which influence the gearshift quality

Model 1„gearshift“

P1 P2 P3 P4

TCU-Parameters

AV

L-D

RIV

E™

Ra

tin

g

A calibration parameter,

which does have a big

influence on the shift quality



CAMEO generates best calibration settings based upon target function &

boundary conditions.

Example: Target Function

Example: Constraints

Boundary conditions

(constraints)

Optimization loop

over time

Modeling and Optimization Phase Parameter Optimization



Test Plan


Desktop



Desktop

Test Plan

Test Run

Data

Check

Modeling &

Optimization

Map

Generation




Generate your mapsDesktop

Calibration ProcessMap Generation Phase



Map Generation PhaseCalibration variants generated without further measurements required

Calibration Engineer smoothes the tables and generates the optimized

parameters e.g. for Shift Quality using CAMEO Transmission

Sport

Comfort

Other



Model Based Approach allows us to decouple the TestingPhase (vehicle) from the Calibration Phase (office)

Test Plan


Desktop



Desktop

Verification

Test Plan

Test Run

Data

Check

Modeling &

Optimization

Map

Generation




Get your mapsDesktop

Release your calibration

Vehicle



Calibration Time CycleRequired Resources for Phases

1st calibration

Test Plan Test Run Modeling Optimization VerifyResources

Calibration Phases

Engineer

Dyno or Powertrain

Vehicle

Get Maps

time duration

~ 4 weeks

*example for one vehicle variant with a 6 speed MT



Calibration Phases

Calibration Time CycleCalibration variants generated without further measurements required

fast optimization loops (customer input)

1st calibration

further calibrations

Test Plan Test Run Modeling Optimization VerifyResources

Calibration Phases

Engineer

Dyno or Powertrain

Vehicle

Get Maps

Resources

Engineer

Dyno or Powertrain

Vehicle

Test Plan Test Run Modeling Optimization Verify

Calibration Phases

Get Maps

Repeat Process

time duration

~ 4 weeks

*example for one vehicle variant with a 6 speed MT

time duration

1-2 days



Calibration Time CycleCalibration variants generated without further measurements required

Verification

Test Plan

Data

Check

Modeling &

Optimization

Map

Generation




Get your mapsDesktop

Release your calibration

Vehicle

fast optimization loops (customer input)further calibrations



Summary:

Time saving of 25%

per vehicle variant

Reduction in vehicle usage

of 30% per vehicle variant

ACT™ represents a new

automated calibration

methodology

We automated the process

We employed a global model-based approach with objective driveability rating

We developed calibration tools, methodologies and processes

1. We are not able to cover all

potential calibration variants within

the available time frame…

2. We have to deliver variant

calibrations with fewer and fewer

prototype vehicles…

3. Our calibration methodology needs

to change to address these critical

issues…



Driveability calibration (mode change) of hybrid powertrain

on the chassis dyno

Battery Simulator

Brake Pedal Actuator



50 calibration engineers in the team

Well connected worldwide

Customers are OEMs and Tier 1 all over the world

AVL offers

Transmission, Hybrid & e-Vehicle

Calibration Programs up to production level

AVL TRANSMISSION CALIBRATION

HYBRID CALIBRATION

e-

e-VEHICLE CALIBRATIONTRANSMISSION CALIBRATION



Gerhard KokaljHead of Transmission,

Hybrid & e-Vehicle Calibration

[email protected]

+43 664 827 43 43

AVL List GmbH Graz, Austria

READY FOR THE ROADFull vehicle calibration up to production level by using AVL technology

mailto:[email protected]

significant reduction of powertrain calibration effort in production programs avl en

Documents

chinagerhard kokalj

avl list gmbh

calibration tools

number of parameters

avl cameoi

test plan phaseobjective

calculation of test

weekend test