FCP Short CourseData Validation and Reduction for Fatigue
Stephen D. DowningMechanical Science and Engineering
© 2001 - 2015 University of Illinois Board of Trustees, All Rights Reserved
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 1 of 29
Outline
Test PlanningData AcquisitionData ValidationData Reduction Modes
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 2 of 29
Data Collection Exercise
Vehicle Instrumentation
Test Planning
Data Collection
Data Validation
Data Analysis
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 3 of 29
Why Do We Collect Data?
Assess product performance Estimate product durability Establish customer usage Model Verification Put Out Fires Other
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 4 of 29
Data Collection Equipment
Analog Tape RecordersDigital Tape Recorders Solid State Digital DevicesData Loggers Time History Recorders Field Computers
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 5 of 29
Data Collection Equipment
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 6 of 29
Field Data Collection
Field tests are expensive Do it in one trip Do I have the right data? Do I have enough to be representative? Is it of sufficient quality and accuracy?
Good data acquisition equipment Rugged Easy to use Flexible
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 7 of 29
Huge Investment
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 8 of 29
Durability Testing
Test Article
Strain Gages
Data Collector
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 9 of 29
What Is Data Validation?
Process of verifying that the data collected meets the objectives set forth during test planning
Data quality is sufficiently high Data quantity is enough to ensure statisical
relevancy Data can sometimes be corrected
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 10 of 29
What Kind of Effort Is Validation?
Most laborious, least interesting form of data analysis
Time consuming - 20% to 50% of total data analysis effort
Critically important - bad data begats bad analysis
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 11 of 29
Data Validation Steps
Visual Inspection
Quality Tests
Sufficiency Tests
Corrective Action
Data Transfer
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 12 of 29
Outline
Time History Time at Level Histograms Sequential Peak Valley Peak Valley Matrix (Markov)Rainflow Matrix FFT
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 13 of 29
Time History
Burst History
Time at Level
Peak-Valley
Time HistoryRange - Mean To - FromRange Only
Rainflow/ Peak-Valley
Matrix
Relationship of Modes
FFT
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 14 of 29
Time History - everythingBurst History - finding rare eventsSequential Peak Valley - durability analysisTime @ Level Histogram - usage and performance
determinationPeak Valley Matrix - statistical history characterizationRainflow - durability analysis and statistical history
characterizationFFT –everything like a time history but in the frequency
domain
Data Mode Uses
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 15 of 29
Time History Maintains amplitude Maintains sequence Maintains phase between multiple channels Maintains frequency content
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 16 of 29
Burst History (Triggered) Maintains amplitude Maintains sequence Maintains phase between multiple channels Maintains frequency content
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 17 of 29
0 5.49time-10
10
Load
Peak Valley Extraction
Maintains amplitude Maintains sequence Destroys phase between multiple channels Destroys frequency content
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 18 of 29
Time At Level Histogram Maintains statistics Destroys sequence Destroys phase between multiple channels Destroys frequency content
Counts the total number of samples in each input signal category.0
0 200-100
100
2
8910
76543
1
7
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 19 of 29
0 1.4Collection Time (secs)-400
400
Volts
-400 400Volts0
79
Counts
Counts the total number of samples in each input signal category.
1D Time At Level
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 20 of 29
0 200Time-100
100
Sine
0 200Time-100
100
Cosine
-100
Sine
-100 100Cosine
100
A
A
B
B
2D Time At LevelCounts the total number of samples in each unique combination of input signal categories
Maintains statistics Destroys sequence Maintains time between
multiple channels Destroys frequency content
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 21 of 29
Counts the total number of samples in each unique combination of input signal categories
Cosine
0 200Time-100
100
-100
100
Sine
-100
100
Sine
-100
100
Cosine
0
2
Cou
nts
2D Time At Level
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 22 of 29
0 5.49time-10
10
1
2
8
9
10
7
65
4
3
B
C
D
E
F
G
H
RANGE
RANGE = | A - B | MEAN = ( A + B ) / 2
PEAK VALLEY PAIRS = AB, BC, CD, DE, EF, FG, GH, HI, IJ, …...
MEAN J
A
Peak Valley Matrix
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 23 of 29
0 27Scans-400
400
0 800Range (uStrain)0
18
Counts
Peak Valley Matrix (Range Only)
Maintains amplitudes Maintains sequence Destroys phase between
multiple channels Destroys frequency content
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 24 of 29
0 27Scans-400
400
RangeMean 50 150 250 350 450 550 650 750-350-250 1-150 1 1 1-50 5 150 6 1150 6 3 1250350
Peak Valley Matrix (Range-Mean)
Maintains amplitudes Maintains sequence Destroys phase between
multiple channels Destroys frequency content
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 25 of 29
0 27Scans-400
400
FROMTo -350 -250 -150 -50 50 150 250 350-350 1-250 1-150 1-50 1 1 4 150 3 6 1150 1 5250 1350
Peak Valley Matrix (To-From)
Maintains amplitudes Maintains sequence Destroys phase between
multiple channels Destroys frequency content
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 26 of 29
Rainflow Counting
Maintains amplitudes Maintains sequence Destroys phase between
multiple channels Destroys frequency content
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 27 of 29
0 5.49time-10
10
1
2
8
9
10
7
65
4
3
A
B
C
D
E
F
G
H
RANGE
PEAK VALLEY PAIRS = AB, BC, CD, DE, EF, FG, GH, HI, IJ, …...
MEAN J
RAINFLOW PAIRS = AB, CD, EF, GH, IJ, …...
Rainflow Counting
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 28 of 29
FFT (How’s Fourier Work?)
Time
Stre
ss
FrequencyFF
T St
ress
Time Domain Frequency Domain
Fourier Transform
Inverse Fourier Transform
ME 431 - Lecture 21 © 2015 Stephen Downing, University of Illinois at Urbana-Champaign, All Rights Reserved 29 of 29
How’s Fourier Work?
Frequency
Am
plitu
de
FourierTransform
etc...
(Theory of Fourier, 1768-1830)
Produced by theCollege of Engineering
University of Illinois at Urbana-Champaign
© 2001 - 2015 University of Illinois Board of Trustees, All Rights Reserved
FCP Short Course
Top Related