Download as pdf or txt
Download as pdf or txt
You are on page 1of 71

Durability Analysis

Detailed Analysis Agenda

• Background: Fatigue Analysis

• Durability Analysis for

– Blockcycle Testing

– Test Acceleration

– Correlation
Fatigue Life Analysis Methods

Total Life Method: S – N l


o
g
(Stress Life/Load Life/Generic Load Life) S

S = AN k log( S ) = log( A) + k log( N )


log N

Crack Initiation Method: e - N ε


(Strain Life) ε = aN b + cN d
log N
Stress-Life Curves

• Generic Stress Life

– Stress life/Load Life


approach
– Specify a Y-axis
intercept and slope
– Amplitude
p to life is linear Amplitude
(for a segment)

– Typically used for A


A-B
B
comparisons
Life
Strain-Life Curves

• Strain-Life
– Data found in lookup
tables e.g. SAE
– Use for true strain data
– Estimates life to crack
initiation
– Units supported
pp are SI,,
USC and KGF
Damage Calculations

• Damage can be calculated from two sources:


– Time history
– Histogram

• Damage can be represented in the following form:


– Damage listing
– Damage Time history
– Damage Histogram
Damage Calculations
• Damage calculation process - 1
1-1. Identify first cycle

l
o
g

log N
1 2 Calculate damage (passes to failure)
1-2.

2-1. Identify next cycle …. Through to last cycle

l
o
g

log N
2-2. Calculate damage (passes to failure)
Damage Calculations

• Damage calculation process – 2

• DamageTotal = (Damagecycle1+Damagecycle2+ …. +DamagecycleN)

• Palmgren-Miner Rule
– Damage per cycle identified from S-N curve
– Total damage is the sum of each cycle
cycle’s
s damage
– Life is inverse of sum of damage
Standard Fatigue Features

The following is a summary for the standard fatigue


analysis features available in Analyze:

– Materials library – Rainflow counting


– Fatigue sensitive editing – Range Pair counting
– Local stress/strain life – Level crossing analysis
prediction – Rosette analysis
– St
Stress
ess life
epprediction
ed ct o – Peak valley slicing
– Mean Stress Correction – Rainflow regeneration
– Damage calculations – Peak valley regeneration
– Notch factor analysis – Histogram accumulation
– Scale factor analysis – Histogram editing
Analyze

• Data formats supported:

– Native file format is RPC format


• This is the defacto-standard in the testing
industry

– Somat SIF files

– nCode DAC files

– ASCII data (time history and matrix)


Agenda

• Background: Fatigue Analysis

• Durability Analysis for

– Blockcycle Testing

– Test Acceleration

– Correlation
Accelerated Durability Testing – Simple Cyclic

Range-Mean
g Material Properties
Rainflow Matrix ε
Measured
εm
Service Strain
εr
N

Calculate fatigue damage

ε
S

Adjust applied load to achieve


t
N selected strain level
Select strain level Cyclic Test
andd cycle
l count Load
Equivalent Damage
• Equivalent Amplitude
– Back calculate a constant
amplitude profile which
yields the same damage as
the time history
– Two modes:
• User defined amplitude
• User
U d
defined
fi d cycles
l
Peak Picking/Peak Valley Slicing

• Peak Picking – Peak Valley Slicing Tool


Modes supported:
Peak Picking
only valid for a single channel test –
phase is not maintained

Peak Slicing
Use for accelerating tests and
compressing data (reduce the size of
data files required for fatigue analysis)

Peak Pick FlexTest Profile


Or
RPC Time History
Accelerated Durability Testing – Block Cycle

Range-Mean
g Material Properties
Rainflow Matrix ε
Measured
εm
Service Strain
ε0
εr
N

εm ε0 ε

εr t S

Remove strain ranges Develop equivalent


t
below fatigue limit block cycle strain history
Block cycle
Adjust applied loads to achieve
test load
selected strain levels
Manual Block Cycle Test Definition
• Histogram Editing – applications
– Component test generation

Export to Excel for further analysis and block cycle definition


Histogram Editing
• Histogram Editor – fatigue sensitive editing

1. Select Rainflow Histogram

2. Configure Damage
3. Calculate Damage
Histogram Editing
• Histogram Editor – fatigue sensitive editing

Black = Rainflow
Blue = Damage

4. Enter editing
criteria & Edit

Fatigue Sensitive Editing will remove


the least damaging content bin-by-bin
until the user specified criteria is met.
Histogram Editing
• Histogram Editor – fatigue sensitive editing

Black = Rainflow - edited


Blue = Damage

5. Output edited Rainflow


Block Cycle Generator

1. Select Rainflow Histogram

2. Review Cyclic and Damage Content


Block Cycle Generator : Zero Bins

1. Define Damage to Retain

2. Run Editing Function

3. Review Resulting Cyclic and Damage Content


Block Cycle Generator : User Defined Bins

1. Define Bins

2. Run Editing Function

3. Review Resulting Cyclic and Damage Content


Block Cycle Generator : Create Profile

1. Define Ordering
2. Define Output Type
3. Run the creation step

4. Review Resulting Profile


Rainflow Regeneration
• Histogram Editing – applications
– Component test generation from edited rainflow
histograms – Rainflow Regeneration
Only applicable for
single channel tests

Export to Excel for further analysis and block cycle definition


Agenda

• Background: Fatigue Analysis

• Durability Analysis for

– Blockcycle Testing

– Test Acceleration

– Correlation
Why Edit the Measured Data?
• Goal of simulation: reproduce the damage caused by
the road in a controlled environment.
• Goal of editing: accelerate test

C t
Customer Usage
U
Why Edit? :

• Shorten test time

• Remove non-
damaging data
Test Track
• Remove
physically
unrealizable data

T Lab
Test L b
Selection of Test Profiles

• Test Profile Selection


– Customer Correlation
• Ultimately the goal of any test
is to test a sample of products
to a customer correlated life
• Several
S l Ch
Challenges
ll exist:
i
– Many different types of
customer
– Many different driving
environments
• Pave or Belgian Block
• Washboard
– Proving Ground events • Pot Holes
• Not all events will be • Curb Strikes
meaningful for all components • Rough Road
• Cross Country
• Ride & Handling
– Need a methodology for selecting
• Other Special Events
the most representative tests
Selection of Test Profiles

• Test Profile Selection

– Damage Assessment Analysis


• RPC Test Profile selection wizards
– Facilitate selection of events before file based
editing
– Damage and time criteria

– Purpose of analysis:
• Allow user to modify test sequence contents
(profiles and passes) to retain damage but reduce
time
• Report correlation of edited data to raw test
• Report correlation of simulated data to raw test
Selection of Test Profiles –
Damage Assessment Analysis
• Results
Simulated Data
Individual Damage
acc_fw_1_twist_1_R shock_1_Rripple_1_Racc_bk_1_inner_1_Rcob_1_RSScale
1EEngBrktVert
B ktV t .00006
00006 .00008
00008 .0094
0094 2 25
2.25
2 EngBrktLat .00186 .00341 .0152 60.
3 EngBrktLng .00379 .00831 .00335 2.375
All Channels .00572 .0118 .02795
Time Original Raw Simulated
Data acc_fw_1_twist_1_R shock_1_Rripple_1_Racc_bk_1_inner_1_Rcob_1_RSCumulative Timee
Time (sec) 300 625 2520 3445
Time Contribution 8.71% 18.14% 73.15% Comparison
Deleted 71.43% 83.33% 62.5% 78.2%
Cumulative Damage to Raw
Edited Data acc_fw_1_twist_1_R shock_1_Rripple_1_Racc_bk_1_inner_1_Rcob_1_RSCumulative
1 EngBrktVert .00193 .00201 .78955 .79349
2 EngBrktLat .0558 .0853 1.27716 1.41827
3 EngBrktLng .11381 .20772 .28136 .60289
All Channels .17154 .29503 2.34807 2.81464
Simulated
Cumulative Contribution
Data acc_fw_1_twist_1_R shock_1_Rripple_1_Racc_bk_1_inner_1_Rcob_1_RSP_12ch_sim
1 EngBrktVert 0.24% 0.25% 99.5%
2 EngBrktLat 3.93% 6.01% 90.05%
3 EngBrktLng 18.88% 34.45% 46.67%
All Channels 6.09% 10.48% 83.42% Simulated
Retained Damage
acc_fw_1_twist_1_R shock_1_Rripple_1_Racc_bk_1_inner_1_Rcob_1_RSCumulative
Damage
Damage
1 EngBrktVert 70.22% 67.89% 81.66% 80.74% Comparison
Retention
2 EngBrktLat 126.18% 126.18% 120.03% 120.12%
to Raw
3 EngBrktLng 61.24% 58.14% 56.17% 57.39%
All Channels 73.68% 68.96% 92.74% 87.57%
Max 126.18% 126.18% 120.03% 120.12%
Min 61.24% 58.14% 56.17% 57.39%
Avg 85.88% 84.07% 85.95% 86.08%
Selection of Test Profiles –
Damage Assessment Analysis

• Results

Cumulative Rainflow Cumulative Damage


(Exceedance Plot) (Summation Plot)
Data Editing

• RPC offers the following editing methods

– Graphical / Manual editing

– Statistical based editing

– Fatigue based editing


Graphical Editing
• Graphical Editing Procedure
– Graphically select the data you wish to edit using
the mouse (user selections are illustrated in red)

– Many
M techniques
t h i supported
t d ffor removing
i and
d re-
joining data at edit boundaries.
Advantages and Disadvantages

• Graphical Editing

– Advantages:
• Interactive – user has full control
• Fast on small data files
• Easily undo/redo edits

– Disadvantages:
• Interactive – cannot readily automate
• Takes more time on larger, more complex
datasets
• Difficult to document editing criteria
Statistical Editing Tools

• Perform statistical selections


on a time historyy

– Mark regions (sections of time


history) to edit based on
Statistical criteria
– Use criteria based on Max/Min,
RMS, Damage, etc
– The marking limits can either
be defined as percentage of
full scale or in absolute values.
– Optionally
O ti ll use multiple
lti l criteria
it i
on different channels with
logical operators
Statistical Editing Example

• Example with three channels of strain data. (Note this


technique
q can be applied
pp to any
y time historyy data.))

- Raw Data
Statistical Editing Example

• Statistical Editor Wizard

1. User enters input data file &


defines output

1. Select statistic and comparison.

3. Define the threshold and if it is


engineering unit or percent.

4. Define window size.

In the example configuration above we are attempting to:


− select 1 second windows
− where the max value within that windows is less than 30% of the overall max
Statistical Editing Example

• Individual Channel Results:

− Note: the regions


g ((the blue sections)) have been selected different for each channel.
− To preserve phase in multi channel data we must combine the independent regions.
Statistical Editing Example

• Combined Region Result:

− After combining the individual channels into a time line region file,
file we can edit the data
data.
− The next figure shows the edited output compared against the unedited input.
Statistical Editing Example

• Edited Result:

Channel 1

Channel 2

Channel 3

Black: Raw unedited data


Bl
Blue: Edit
Edited
dddata
t
Advantages and Disadvantages

• Statistical Editing

– Advantages:
• Automation
• Easy/quick to edit even larger data files
• Can combine multiple editing criteria
• Can manually “fine
fine tune”
tune results
• Consistent editing procedures can be defined

– Disadvantages:
• May not be intuitive
• Additional
Additi l analysis
l i may be
b required
i d tto
achieve time reduction goals
Damage Time History

• Damage is calculated for every peak-valley cycle, the


damageg time history y is g
generate by
y assigning
g g damageg
to location of the peak and valley.

Peak – valley pair

Peak – valley pair


Damage Time History

• Fatigue Sensitive Editing:

Edit time history data based Damage time history


on regions selected from the shows where damaging
damage time history cycles occur in the data
Fatigue Sensitive Editing

• Automatic Damage Editor (ADE)

– The ADE tool is a fatigue sensitive editor and


allows the user to accurately and automatically
reduce
d th
the llength
th off d
data
t fil
files required
i d ffor
simulation.

– By specifying a desired damage retention


percentage, the tool will iterate to an optimum
solution balancing damage retention with time
reduction.

– The
Th ADE lleverages th
the concepts t off D
Damage
Time History and Statistical Editing
Fatigue Sensitive Editing

y ADE Tool –
editing
configuration

Use the Automatic


g tab to
Editing
specify a desired
percentage of
damage retention.
Fatigue Sensitive Editing

• Automatic Damage Editor Tool – execution


window

tolerance

- The damage time history (blue) represents the per cycle occurrence of damage
in the strain road data (black)
- Regions are selected by statistically processing the damage time history and
p g against
comparing g a tolerance value
- If the damage is less than the tolerance the window is selected for deletion
- The tolerance value is automatically adjusted to “iterate” to a solution
Fatigue Sensitive Editing

• Automatic Damage Editor Tool – reporting results

The report is
automatically
updated every
pass of the ADE
tool.
Fatigue Sensitive Editing

• Automatic Damage Editor Tool – reporting results

Color highlight
g g is
used to indicate
the progress of
each channel, red
indicates
channels out of
criteria while
green indicated
those channels
that are within
criteria.
Fatigue Sensitive Editing

• Automatic Damage Editor Tool – selected regions

Regions
marked
for
deletion
Fatigue Sensitive Editing

• Automatic Damage Editor Tool – final output

Raw and Edited Time History


Fatigue Sensitive Editing - Example

• Objective: to perform fatigue sensitive editing on data


provided by
p y a customer to establish total durability
y
test time and the amount of feasible test time
reduction while maintaining adequate fatigue content
of the raw road data.

– Four test cases are evaluated:


• Analysis of original customer editing (RMS
based)
• 99% fatigue retention
• 95% fatigue retention
• 20 hour test
Fatigue Sensitive Editing - Example

Result Summary

T t Time
Test Ti Reduction
R d ti

% Original Test
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Raw Test

Statistical
St ti ti l
Editing

99% Damage
Retention

95% Damage
Retention
Fatigue Sensitive Editing - Example
• Customer Editing (Test time = 183.1 hrs Time Reduction = 68%)
Ca se : Custom e r Editing

Damage Cha nne l Ra w da m a ge Edite d da m a ge Re te ntion


Analysis X 4.52E-02 4.49E-02 99.3%
Y 2.31E-02 2.30E-02 99.4%
Z 1.37E-01 1.31E-01 95.7%

Ave ra ge Re te ntion 98.1%

Levelcross
of raw and
edited data

Black = Raw
Blue = Edit
Fatigue Sensitive Editing - Example
• 99% damage retention (Test time = 151.3 hrs Time Reduction
= 74%) Ca se : 99% Fa tigue Re te ntion
tollerence = +/- 0.5%
Damage Cha nne l Ra w da m a ge Edite d da m a ge Re te ntion
Analysis X 4.52E-02 4.47E-02 98.9%
Y 2.31E-02 2.29E-02 98.9%
Z 1.37E-01 1.35E-01 98.8%

Ave ra ge Re te ntion 98.9%

Levelcross
of raw and
edited data

Black = Raw
Blue = Edit
Fatigue Sensitive Editing - Example
• 95% damage retention (Test time = 122.8 hrs Time Reduction
= 79%)Ca se : 95% fa tigue re te ntion
tollerence = +/- 2%
Damage Cha nne l Ra w da m a ge Edite d da m a ge Re te ntion
Analysis X 4.52E-02 4.37E-02 96.7%
Y 2.31E-02 2.19E-02 95.0%
Z 1.37E-01 1.30E-01 95.2%

Ave ra ge Re te ntion 95.6%

Levelcross
of raw and
edited data

Black = Raw
Blue = Edit
Fatigue Sensitive Editing - Example
• 20 Hour Test (Test time = 22 hrs, Time Reduction = 96%, Fatigue
Retention = 29%)
Case: 20hr Test

Damage Channel Raw damage Edited damage Retention


Analysis X 4.52E-02 1.43E-02 31.6%
Y 2.31E-02 6.69E-03 28.9%
Z 1.37E-01 3.69E-02 26.9%

Average Retention 29.1%

Levelcross
of raw and
edited data

Black = Raw Note: This level of


Blue = Edit test acceleration was
unacceptable because
of the poor fatigue
retention.
Fatigue Sensitive Editing - Example
• Comparison of retained damage – histogram
analysis (Raw)
Fatigue Sensitive Editing - Example
• Comparison of retained damage – histogram
analysis (99%)
Fatigue Sensitive Editing - Example
• Comparison of retained damage – histogram
analysis (20Hr)
Fatigue Sensitive Editing

• Fatigue Sensitive Editing Summary

– Fatigue Sensitive Editing is an alternative editing


approach which is accurate and efficient.

– The Automatic Damage Editor allows users


understand the trade-off of test time reduction and
damage retention.
Agenda

• Background: Fatigue Analysis

• Durability Analysis for

– Block Cycle Testing

– Test Acceleration

– Correlation
Why Validate the Test Results?
• Goal of simulation: reproduce the damage caused by the road
in a controlled environment.
• Goal of validation: Prove that y
you have replicated
p the test track.

C t
Customer Usage
U
Why Validate? :

• Defend your test


results.

• Learn where
Test Track improvements
are needed.

• Gain confidence
=? in accuracy of
models.

T Lab
Test L b
Basis for Correlation

• Correlation Sensors
– Sensors (load, strain, acceleration, etc) which are
measured but are not used as part of the Control Set

• Approach
– Very similar to test acceleration

• Tools For Correlation


– Visual Inspection
p
– Rainflow Analysis
– Fatigue Analysis
– Frequency Analysis
– Statistical Analysis
– Level Crossing Analysis
Basis for Correlation

• test_timehistory: measured data from test


• target_timehistory:
target timehistory: “desired”
desired data from PG
Visual Inspection
Visual Inspection
Rainflow Analysis
Damage Assessment Analysis

• Results
Simulated Data
Individual Damage
acc_fw_1_twist_1_R shock_1_Rripple_1_Racc_bk_1_inner_1_Rcob_1_RSScale
1EEngBrktVert
B ktV t .00006
00006 .00008
00008 .0094
0094 2 25
2.25
2 EngBrktLat .00186 .00341 .0152 60.
3 EngBrktLng .00379 .00831 .00335 2.375
All Channels .00572 .0118 .02795
Time Original Raw Simulated
Data acc_fw_1_twist_1_R shock_1_Rripple_1_Racc_bk_1_inner_1_Rcob_1_RSCumulative Timee
Time (sec) 300 625 2520 3445
Time Contribution 8.71% 18.14% 73.15% Comparison
Deleted 71.43% 83.33% 62.5% 78.2%
Cumulative Damage to Raw
Edited Dataacc_fw_1_twist_1_R shock_1_Rripple_1_Racc_bk_1_inner_1_Rcob_1_RSCumulative
1 EngBrktVert .00193 .00201 .78955 .79349
2 EngBrktLat .0558 .0853 1.27716 1.41827
3 EngBrktLng .11381 .20772 .28136 .60289
All Channels .17154 .29503 2.34807 2.81464
Simulated
Cumulative Contribution
Data acc_fw_1_twist_1_R shock_1_Rripple_1_Racc_bk_1_inner_1_Rcob_1_RSP_12ch_sim
1 EngBrktVert 0.24% 0.25% 99.5%
2 EngBrktLat 3.93% 6.01% 90.05%
3 EngBrktLng 18.88% 34.45% 46.67%
All Channels 6.09% 10.48% 83.42% Simulated
Retained Damage
acc_fw_1_twist_1_R shock_1_Rripple_1_Racc_bk_1_inner_1_Rcob_1_RSCumulative
Damage
Damage
1 EngBrktVert 70.22% 67.89% 81.66% 80.74% Comparison
Retention
2 EngBrktLat 126.18% 126.18% 120.03% 120.12%
to Raw
3 EngBrktLng 61.24% 58.14% 56.17% 57.39%
All Channels 73.68% 68.96% 92.74% 87.57%
Max 126.18% 126.18% 120.03% 120.12%
Min 61.24% 58.14% 56.17% 57.39%
Avg 85.88% 84.07% 85.95% 86.08%
Frequency Analysis
M

-15
-10
10
-5
0
5
10
15
ax
im
um

M
in
im
um

M
ea
n

V
ar
ia
nc
e

S
Statistical Analysis

td
D
ev

R
M
S

C
re
st

S
um

K
ur
to
si
s

S
ke
w
Test
g
Target
Level Cross Analysis

• How many times does the signal cross each level?


RPC Pro – Reporting

• Time History Report Tool


– Template based Report
Generation
– Fully customizable

You might also like