Using Fatigue Damage Spectrum for Accelerated Testing with
Transcription
Using Fatigue Damage Spectrum for Accelerated Testing with
Automotive Testing Expo 2016 Stuttgart Germany 31th May - 02th June 2016 Using Fatigue Damage Spectrum for Accelerated Testing with Correlation to End-Use Environment Lecturer: Holger Boller Vibration Research About Vibration Research § § § § § Vibration Research formed as a private company in 1995 45 employees in development, research and production More as 4.500 installed Vibration Control system world wide Global Customer Base In 2014 new building in Jenison § Pioneered many vibration testing techniques § First to perform long time history replication on ED shakers - We call it F.D.R. Field Data Replication § New featuring kurtosis control (Kurtosion®) § VRC has received U.S. patent number 4,726,726 for its Kurtosion® software § Introduced FDS - Fatigue Damage Spectrum § Introduced iDOF - Instant Degrees of Freedom Using Fatigue Damage Spectrum for Accelerated Testing with Correlation to End-Use Environment § Key Point´s and Questions for the customer are: 1. Can I reproduce the measured different complex time signals in an easy way? 2. How can I generate PSD profiles from the time signals in a realistic and reproducible way? 3. Comparison of the fatigue damage between the Time signal and the generated RANDOM signal Actual Situation - Where are we? Actual Situation § Simulation of mechanical stress is included in the design, development and qualification of products § Use of mathematically generated time signals § Derivation of parameterized models § Good correspondence between failures in the field and in the laboratory simulation None of the signals thus generated is identical to the load in the field. We can´t reproduce and justify all failures. Vibration Testing in the Real World § Use of standardized test specifications § ISO 16750; MIL 810; RTCA DO160; VW 80000; ASTM 4728; EN 60068; IEC 17025 § Can´t answer all Questions § A lot of possibilities and an a big field of interpretation § Many papers to read Handling of the Time Signal § Create a Test Profile from Time History Data § § § § Method 1: Peak Hold PSD Method 2: Average PSD Method 3: Field Data Replication Method 4: PSD Profile with the same Fatigue Method 1 - Peak Hold PSD § PEAK HOLD - FFT based on peaks, from each time block AccelerationSpectralDensity(G²/Hz) § Extracts the peaks from the blocks of data § Max Values from each Time Block are used to create the PSD Spectrum § Critical for time signals with short events (pulses) AccelerationSpectralDensity Demand 1x100 1x10-1 1x10-2 1x10-3 1x10-4 1x10-5 1x10-6 5 10 Frequency(Hz) 100 200 Method 2 – Average PSD § AVERAGE - FFT of each block of data is averaged together AccelerationSpectralDensity(G²/Hz) § Creates a uniform average - power of the test § Critical for time signals with highly variable RMS value § This method “lose” short events AccelerationSpectralDensity Demand 1x100 1x10-1 1x10-2 1x10-3 1x10-4 1x10-5 1x10-6 5 10 Frequency(Hz) 100 200 Method 3 – Field Data Replication § Exact replication of the Time History – Amplitude, Phase, Fatigue Cycles § Maintains ordering of Stress Cycles § Cannot be easily accelerated Comparison between PEAK HOLD PSD, AVERAGE , TIME SIGNAL Signal gRMS gPEAK vRMS vPEAK dPEAK KURTOSIS TIME SIGNAL 0,44 -3,9 / 4,7 0,05 -0,25 / 0,27 -6,2 / 7,4 ~6 PEAK HOLD PSD 1,86 -11,1 / 11,1 0,21 -1,3 / 1,3 -36,6 / 25,7 ~6 PEAK HOLD PSD 1,87 -5,9 / 5,8 0,21 -0,75 / 0,73 -21,1 / 19,3 ~3 AVERAGE PSD 0,44 -2,6 / 2,6 0,05 -0,39 / 0,31 -7,9 / 7,2 ~6 4 Key Questions came up with our comparison § Is it possible to design a vibration test (profile and test time) that accurately simulates the End-Use Environment of a product throughout its lifetime? § Is it possible to confidently combine multiple, complex time signals into a single test profile? § Is it possible to reliably accelerate a test and maintain an accurate simulation of imported field data? § How I can compare all my new test profiles? Method 4: Fatigue Damage Spectrum § Same recording duration - same test time § Same Frequency Range § Accounts for time history, duration, and fatigue damage TimeSignal AVERAGEPSD PEAKHOLDK=3 PEAKHOLDK=6 1.00E+08 FATIGUEDAMAGE 1.00E+05 1.00E+02 1.00E-01 1.00E-04 1.00E-07 1.00E-10 1.00E-13 1.00E-16 1.00E-19 3 30 Frequency[Hz] 300 Theory Fatigue Damage Calculation § S-N Curve (S for stress, N for cycles): § Miner’s Rule: time strength N = c ∗ S −b permanently § N=Numberofstresscyclestofailure § S=peakamplitudeofcyclicalstress § b=faQgueparameter § c is experimentally found to be between 0.7 and 2.2. Usually for design purposes, c is assumed to be 1. k ni =c ∑ i =1 N i How works the Calculation of the Fatigue Damage Spectrum? § The acceleration waveform is converted to a velocity waveform § Velocity is desired because Henderson-Piersol equations utilize velocity - Velocity has a direct relationship to stress § Stress waveform is narrow-band filtered and the defined Q value is used § Q value determines filter bandwidth § Cycle counting and damage calculation ensure in each frequency-bin to determine damage contributions over a spectrum of frequencies. § Spectrum spacing determines number of points in spectrum (number of frequency bins) How works the Calculation of the Fatigue Damage Spectrum? b Fatigue and c Factor Acceleration Waveform 1 Integrator Acceleration Waveform 2 Integrator Acceleration Waveform 3 Integrator Velocity Waveform 1-3 Multiply Narrow Band Filter Q-Factor • • • Rainflow Counting • • • Fatigue Damage Calculation Henderson Piersol Conversion Kurtosion™ PSD Spectrum How works the Calculation of the Fatigue Damage Spectrum? Fatigue Monitor - VibrationView Software § We have 4 different time signals and would like to make a random test that recreates the same fatigue. § We want to run an equivalent of 10.000 hours of desired life time in 30 hours vibration test § select a ‘b’ and ‘Q’ value for your data and product. −b § “b” is calculated with: N i = c ∗ Si § “Q” or quality factor is used in narrow band pass filter calculations, and will affect the smoothness of the Fatigue damage § If you are not sure then MIL-Std 810 suggests b = 8 and Q = 50. (see pg. 288 of MIL-STD 810G) Creating a RANDOM Vibration Test with the Fatigue Damage Module § § § § § § § Import the Time History file(s) Set the frequency range Enter m and Q Values Enter Target Life Define the Test Duration Add Kurtosis Create the PSD Profile Table Combining FDS and Accelerated Test Time Cobblestone RailroadCrossing ComfortRoad ResonanceRoad Combined 1.00E+07 FATIGUEDAMAGE 1.00E+04 1.00E+01 1.00E-02 1.00E-05 1.00E-08 1.00E-11 3 30 Frequency[Hz] 300 Same Fatigue - without Test Time compression § Weight each waveform by exposure time, and / or number of discrete events § Accelerate the test creating a Gaussian PSD profile that is the damage equivalent to the total target life PSDSpectrum[g²/Hz] Cobblestone RailroadCrossing ComfortRoad ResonanceRoad Combined Frequenz[Hz] 300 1.00E-01 1.00E-03 1.00E-05 3 30 Same Fatigue - with Test Time Compression § Weight each waveform by exposure time, and / or number of discrete events § Accelerate the test creating a Gaussian PSD profile that is the damage equivalent to the total target life Cobblestone RailroadCrossing ComfortRoad ResonanceRoad 30 Frequency[Hz] Combined PSDSpectrum[g²/Hz] 1.00E+07 1.00E+04 1.00E+01 1.00E-02 1.00E-05 1.00E-08 3 300 Conclusions § With the FDS Software Module from Vibration Research: § We have a powerful tool for our customers § To compare given Time Signals in Ratio to the generated Damage § To combine Multiple & Complex measured Time Signals § To create easily a Random PSD Profile that is the Fatigue Damage equivalent to the imported & weighted time history file(s) § To integrate higher Kurtosis Values into the calculated PSD Profiles § Accelerate testing using FDS and Kurtosion® F.D.R Replication Software § Reproduce complex time signals in an easy way § The repeating rate can reflect the life time and the true environmental condition Any Questions? Thank you very much for your attention. Any Questions ?? Come to our Booth: 1268 or Send E-Mail: vrsales@vibrationresearch.com or Visit our website: www.vibrationresearch.com Thank You!