Relating Random Vibration Cycles to Failure to Time

[Cowtree]

Hi guys, I have a simulation I've run a random vibration analysis on for a mounting bracket for an automotive application. I've applied a PSD input to the model and obtained the equivalent stresses for three standard deviations. I've then taken the S-N chart for the martieral and used Miner's Rule to produce a number of cycles to failure for the part.

What I'm curious about is if there is a method to relate those cycles to time. For a periodic input it would be simple to take the input frequency and obtain the time it takes to complete the cycles to failure. I'm not sure where to start with a random input and I'm hoping someone knows of a method or reference for it.

Thanks,
Eric

 

[jimkinney]

I don't have it handy, but Steinberg covers random vibration fatigue in "Vibration Analysis for Electronic Equipment"

Jim

Jim Kinney
Kennedy Space Center, FL

 

[izax1]

When it comes to Random Vibration Fatigue, there are several methods developed to calculate fatigue life.

1. Steinberg
2. Equivalent Narrow Band (Wirsching)
3. Dirlik

1 being the simplest and most crude, 3 being the most advanced.

You would really need to convert the life to amount of damage. Each frequency in the structure will comtribute to total damage. The total damage you can allow, is up to the engineeer to find out. (MTBF)

 

[Cowtree]

Thanks guys.

The example Stienberg uses is a pretty simple object and he just uses the natural frequency to relate the cycles to failure back to time. Its basically what I had followed but then gotten stuck at with trying to relate cycles to time in a system with more than one natural frequency. I think Dirlik is more what I'm looking for, thanks.