Realistic and customer correlated test schedules · Realistic and customer correlated test schedules. ... Rainflow counting ... Rainflow 1 To From Rainflow (symmetric) 2 To From

Unrestricted © Siemens AG 2017 Realize innovation.

Realistic and customercorrelated test schedules

Unrestricted © Siemens AG 20172017.05.22Page 2 Siemens PLM Software

Need for durability engineeringHow to capitalize on these implications ?



Agenda

Loads and damage

Load characterization

Customer correlation

Accelerated testing and analysis

Applications


Agenda

Loads and damage




Applications


Road A

How to understand the fatigue content of loads ?Comparison of two measurements

Road B

Time Time


How to understand the fatigue content of loads ?Comparison of two measurements

Road A Road B

DamageDamage

Time Time


How to understand the fatigue content of loads ?Damage calculation

DamageFatigue life (= 1/Damage)

S-N curve forCA loading

Loads

Failure

t, time

, F

Rainflow countingCycles, range & mean

Damage accumulation


How to understand fatigue content of loads ?What is an S-N curve ?

Finite life Infinite lifeHigh cycle

fatigue(Elastic behavior)

Low cyclefatigue

(Plastic behavior)

Number of stress cycles to failure (N)

Stre

ssam

plitu

de(S

)

1E+00 1E+01 1E+02 1E+03 1E+04 1E+05 1E+06 1E+07 1E+08

N1

S1


10987654321

Classificationinto bins

How to understand fatigue content of loads ?Endo (1968). Rainflow counting for variable loads

Tatsuo Endo*1925 1989

Rainflow counting = technique to decompose a spectrum of varying, complex stressinto a set of simple stress reversals

‘From’loadlevel

‘To’ load level

12345678910

1 2 3 4 5 6 7 8 9 10

Rainflow-matrix:the number of cycles

is cumulated inthe from-to cells

Load history

t

look for closed loops


How to understand fatigue content of loads ?Rainflow characteristics

Zero range

Zero mean

Compression

Tension

Standing

Hanging

9000060000400003000020000100000

t

t0

- mean

t0

+ mean

t

More damage

More damage

To

From

# Cycles


How to understand fatigue content of loads ?Counting methods benefit

• Representations easy to understand• Ability for fatigue related data manipulation (editing, extrapolation, superposition)• Easy way to compare data• Simple counting algorithms• Substantial reduction of data amount

Same counting result = Same fatigue potential

Countingmethod


How to understand fatigue content of loads ?Palmgren (1924) – Miner (1945). Damage accumulation rule

Assume that, during the service life, we have• 500 cycles of load type 1

(defined by mid-value and magnitude)• 1000 cycles of load type 2• and 10000 cycles of load type 3the Palmgren-Miner rule states that failure occurswhen

with:• ni = the number of actual applied load cycles

of type i• Ni = the pertinent fatigue life for that specific

applied load cycle i

= 1

= 0.5 = 0.01 = 0


How to understand fatigue content of loads ?Rainflow Range-pair Damage Cumulative damage

Range-pair (cycle count)

4 0.9 1000.0300.0100.030.010.03.01.0Cycl e Count

184.1

11599.0

1000.0

2000.0

3000.0

4000.0

5000.0

6000.0

7000.0

8000.0

9000.0

10000.0

11000.0

[N]

File Channel Number of bins Lower Limi t Upper Limit Unit Hys teresis FilterD:\Program Files\LMS \TecWare 3.7SL1+\demo\general\nature.erfm FORCE_vert 64 -936.7174 10846.3457 [N] 1

Range Pair

Ran

ge

# Cycles

Cumulative damage

6 0 10010 20 30 40 50 60 70 80 90Cum ulative Damage [100%=6.2541E-3]

184.1

11599.0

1000.0

2000.0

3000.0

4000.0

5000.0

6000.0

7000.0

8000.0

9000.0

10000.0

11000.0

[N]

File Channel Number of b ins Lower Lim it Upper Limit Unit Hysteresis FilterD:\Program Files\LMS\TecWare 3.7SL1+\demo\general\nature.erfm FORCE_vert 64 -936.7174 10846.3457 [N] 1

Range Pair

Ran

ge

Pseudo-damage

Damageaccumulation(Miner’s rule)

Range-pair (damage)

5 0.00000 0.006250.00050 0.00100 0.00150 0.00200 0.00250 0.00300 0.00350 0.00400 0.00450 0.00500 0.00550Damage

184.1

11599.0

1000.0

2000.0

3000.0

4000.0

5000.0

6000.0

7000.0

8000.0

9000.0

10000.0

11000.0

[N]

File Channel Number of bins Lower Limi t Upper Limit Unit Hys teresis FilterD:\Program Files\LMS \TecWare 3.7SL1+\demo\general\nature.erfm FORCE_vert 64 -936.7174 10846.3457 [N] 1

Range Pair

Ran

ge

Damage

Take S-N curveinto account forpseudo-damage

distribution

Axes rotating:X-axis = MeanY-axis = Range

Folding:No distinction

betweenstanding

/hanging cycles

Rainflow (range/mean)

3 Mean

Ran

ge

Rainflow

1 To

From

Rainflow (symmetric)

2 To

From

Only ranges and# cycles are kept.

Mean value isignored


How to understand fatigue content of loads ?Comparison of two measurements

Road A Road B

DamageDamage


How to understand fatigue content of loads ?Scatter in loadings

Scenario: 6 measurementsOne lap on same roadSame weather conditionsSix different drivers

Range-pair histograms (cycle count)

Six different load historiesSix different range-pair histogramsSix different damage valuesSix different fatigue lifes

Damage histograms (cumulative damage)


Agenda

Loads and damage




Applications


How to design realistic test schedule ?Target setting, handling multiple events

1%

Worst customer case

Test track

Special test

Load Cases Load Histories Rainflow Matrices

Extrapolate

Edit

Design


How to design realistic test schedule ?Target setting, handling multiple events

1%

Worst customer case

Test track

Special test

Load Cases

Superposition

10%

25%

65%

TARGET


How to design realistic test schedule ?Target setting – Extrapolation for longer duration

Single lap(measured)

Simply multiplied x6

Six laps(measured)

Extrapolated

xMultiplying

not sufficient!

Six laps ?

Similar to realmeasurement !


How to design realistic test schedule ?Target setting – Extrapolation for longer duration

Original matrix

Extrapolated matrix

Parameterization• Extrapolation factor (“number of laps”)• Smoothing factor (“reliability of data”)

-4-2

02

4-4

-20

24

00.

51

1.5

Compensate for scattering on:Type of drivers | Load data | Component dimensions | Material properties | ...


How to design realistic test schedule ?From target to test procedure

Track mixture

Target

Special test III

Special test II

Special test I

Test track D

Test track C

Test track B

Test track A

cj · track j

Test schedule

Equaldamage !

3x 7x

6x 4x


Target Result

How to design realistic test schedule ?From target to test procedure using LMS Tecware CombiTrack

TargetResult

Target

Nr. of trackrepetitions

TargetdamageDamage

per channel


Agenda

Loads and damage




Applications


Customer correlationStatistical approach Fleet measurement required

LMS CuCo: Fleet measurementTraditional customer correlation

Extensivemeasurementsetup(50+ channels)

Very limitedsample size(1 car, few weeks)

Limitedcustomer

relevance !

Lean CuComeasurementsetup (8 channels)

Extendedsample size(Fleet: 5+ cars,1 year)

Highcustomer

relevance !


Phase 1:Base data collectionon proving ground +

Measurements

Customer correlationCuCo project structure

Phase 4: Analysis

Customer correlated

Phase 2:Correlation model

In Out

Modeling

Phase 3:Fleet measurements


CustomerUsage

DurabilityTest

Pro

babi

lity

Den

sity

Loading / Strength

LMS Customer CorrelationResult: “Fingerprint” of Customer Representative Loading

“Fingerprint” of major vehicle loading components

Braking load (x)1000

100

10

1

Acceleration load (x)Lateral load (y)

Vertical load (z)Low frequency

Vertical load (z)High frequency

1% Customer EUR

Durability Test

1% Customer New Market


Agenda

Loads and damage




Applications


How can you accelerate a test ?

1 Increasespeed

Increaseamplitude2 Omit non-

damaging events3 Simplifythe test4

Basic principle = conservation of damage


How can you accelerate a test ?Increase testing speed

1 Increasespeed




1st naturalfrequency

How can you accelerate a test ?Increase testing speed

Frequency content should stay reasonably below 1st natural frequency (resonances)

Important: Avoid too high compression!

27,4 s13,7 s


How can you accelerate a test ?Increase amplitude

1 Increasespeed




How can you accelerate a test ?Increase amplitude

Logarithmic nature of fatigueChanging slightly the cyclic load appliedto an optimally shaped steel component,has a big influence on the life time of thiscomponent

Load

Number ofcycles to failure

log

log

Life = (load)-k

1

k

+15% ÷2 !

Note:Be careful not to generate an uncharacteristic failuremode.

Yieldstrength

Load (%) Number of cycles (%)

100 100

115 50

87 200


How can you accelerate a test ?Omit non-damaging events

1 Increasespeed




How can you accelerate a test ?Omit non-damaging events – Uni-axial

Acceleration of x14 !100070

All cycles below 4000N only contributeless than 0.5% of the total damage

If you remove these from the loading, you endup with 70 cycles instead of 1000


How can you accelerate a test ?Omit non-damaging events - Multi-axial – RP-filtering

Vertical

Lateral

Longitudinal

Longitudinal

Uni-axial rainflow channel-per-channel is not OKPhase relation lost !

VerticalLateral

Different loads !


How can you accelerate a test ?Omit non-damaging events - Multi-axial – RP-filtering

0 152 4 6 8 10 12 14Zeit [s] 1

-2000

3000

-1500-1000

-5000

5001000150020002500

FOR

CE

_lon

gN

1

-3000

2000

-2500-2000-1500-1000

-5000

50010001500

FOR

CE

_lat

N2

-2000

12000

-600800

220036005000640078009200

10600

FOR

CE

_ver

tN3

Attribute Name UnitC:\Programme\LMS\TW 3.3 QC\demo\general\nature.dmd,FORCE_long NC:\Programme\LMS\TW 3.3 QC\demo\general\nature_edt.dmd,FORCE_long NC:\Programme\LMS\TW 3.3 QC\demo\general\nature.dmd,FORCE_lat NC:\Programme\LMS\TW 3.3 QC\demo\general\nature_edt.dmd,FORCE_lat NC:\Programme\LMS\TW 3.3 QC\demo\general\nature.dmd,FORCE_vert NC:\Programme\LMS\TW 3.3 QC\demo\general\nature_edt.dmd,FORCE_vert N

Small amplitudes/little damage

Large amplitudes/high damage

0.0 27.75.0 10.0 15.0 20.0 25.0Zeit [s] 1

-3510.1

3510.1

-3000.0-2000.0-1000.0

0.01000.02000.03000.0

FOR

CE_

long

N1

-2062.6

2062.6

-1500.0-1000.0-500.0

0.0500.0

1000.01500.0

FORC

E_l

atN

2

-10509.7

10509.7

-7500.0-5000.0-2500.0

0.02500.05000.07500.0

FOR

CE_v

ertN

3

At tribute Name UnitC:\Programme\LMS\TW 3.3 QC\demo\general\nature.dmd,FORCE_long NC:\Programme\LMS\TW 3.3 QC\demo\general\nature_edt_edt .dmd,FORCE_long NC:\Programme\LMS\TW 3.3 QC\demo\general\nature.dmd,FORCE_lat NC:\Programme\LMS\TW 3.3 QC\demo\general\nature_edt_edt .dmd,FORCE_lat NC:\Programme\LMS\TW 3.3 QC\demo\general\nature.dmd,FORCE_v ert NC:\Programme\LMS\TW 3.3 QC\demo\general\nature_edt_edt .dmd,FORCE_v ert N

Phase relation kept !


How can you accelerate a test ?Simplify the test

1 Increasespeed




How can you accelerate a test ?Simplify the test – Constant amplitude test

Testing withmaximumamplitude

4x moredamaging !

10025

Accelerate by 4xIncrease sine wave

frequency ?


How can you accelerate a test ?Simplify the test – Block cycle test

Mix of different ‘Constant-amplitude’ tests for more representative results

Original time series

Normal/incidental/accidental(automatic or user-defined)

Damage calculation(Standard S-N curve or user-defined)

Rainflow matrix

Block cycle test

Damage equivalent 3 3612


Agenda

Loads and damage




Applications


Accelerated durability testing cycles• Meeting 1.2 million km durability requirement• Real tests would take 3 years

• Large-scale customer data collection• 5000 km Turkish public road data• Ford Lommel proving ground

• Development of accelerated rig test• Target setting• Test schedule definition

• Resulting test schedule 8 weeks• Test acceleration of factor 100

Application case Ford OtosanDriving 1.2 million kilometers in 8 weeks

LMS engineers performed dedicated data collection, applied extensive load data processingtechniques and developed a 6- to 8-week test track sequence and 4-week accelerated rig testscenario that matched the fatigue damage generated by 1.2 million km of road driving.


Application case Ford OtosanProject steps

Validation

1. PVG test2. 4-poster test of the cabine3. 4-poster virtual test

of the cabine

Test scheduledefinition

• Data consolidation• Data analysis• Target 1 Mio km

Goal• 10.000 km PVG

Durability test

Data collection

• 4 months on Turkish roads• 1 week proving ground

Preparation

Loads definition• Target vehicle• Current usage (full & empty)

Route selection• 5.000 km• 16 routes with 140 sections

in totalInstrumentation• +/- 60 channels (acc., strain, displ.)


Simcenter durability solutions throughout the development process


Thank you! Want to know more?

• Text• Text• Text




Read more Explore, share and learn

Watch videos Contact the expert

Documents

Realistic and customer correlated test schedules · Realistic and customer correlated test schedules. ... Rainflow counting ... Rainflow 1 To From Rainflow (symmetric) 2 To From