Random Vibration RMS Stress

[mechanicalFEA]

Hi all,

Not sure if this is the best place for this post but here we go...

How do I account for a factor of safety with RMS stresses abtained from an random vibration FEA analysis using an acceleration Vs frequency PSD excitation input. Is it a simple case of doubling the RMS stress value for a factor of safty of 2? ,I plan to use this RMS stress to calculate fatigue life/damage using minors rule but when it comes to the 2 or 3 sigma the already doubled one sigma value is effectivly doubled and then doubled again which surely cannot be correct!?

or

is the damage life prediction adjusted, i.e if a part had a life of 1000 hours at factor of saftey of 1 it would only last 500 hours at FOS 2



Any help, comments or discussion on this topic is welcomed and appreciated,

Many thanks

 

[GregLocock]

If you apply a FoS of 2 to stresses then for a typical steel the thing will last 10 times as long. That may be wise for a proto but seems excessive for a mass produced item.

On the other hand measured stresses vary by 30% or so for supposedly identical events, in my field, so a margin of 20% or so, for a doubling of life, may not be conservative.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[izax1]

Random Vibration is all based on Statistics. Calculating the vibration fatigue life is always a combination of stress level and Nof Cycles at that level. One sigma is the rms level (assuming 0 mean) and probability of reaching the one sigma level is statistically happening at 68,3% of the total time, reaching 2 sigma level is 27,1% of the total time and reaching 3 sigma level is only 4,3% of the total time. So number of cycles is significantly lower for teh higher stress levels. this is what you have to put into your Miners calculations

 

[mechanicalFEA]

Thanks for your inputs, i understand the process to calculate percfent probability and life/cycles etc, after thinking about this i think its clear that FOS is simply calculated by adjusting time to failure, i.e a product must last a duration of a test say 1 hour 10000 cycles = FOS 1. To ensure the product has FOS = 2 it must last for 2 hours at 2000 cycles.

 

[izax1]

I would think that in the case of Random Vibrtion fatigue, you would rather calculate damage rather than FoS. As long as you know how many cycles (or hours) your system has to last, you can convert that to damage accumulated for 1 sigma 2 sigma and 3 sigma stress levels and sum up to calculate total damage.