AnsweredAssumed Answered

Cannot recreate failure at 54 Hz Dwell

Question asked by Travis Perullo on Jul 10, 2014
Latest reply on Jul 15, 2014 by Jared Conway

Like • Show 0 Likes0
Comment • 6

Hello community,

I design aircraft fuel probes, and a design has mechanically failed on the vibration table during a half hour long 54 Hz dwell.

I am attempting to recreate this event in Solidworks Simulation so that we can verify our new design by analysis before going to the machine shop.

I have been trying to do this by first running a Dynamic study that I plan on importing into a Fatigue study.

Since it is a half hour test, I've created a curve with high level spikes (~1.6 m/s^2) from 0 to 97200.

Time Range: 0.01 to 92700, time increment 10 (I have tried much smaller, but it takes forever and inevitably fails).

I am experiencing a few different errors consistently.

During the dynamics calculation I'm often met with "Reaction Node not found in *.lcp file". I have no idea what this means, and can't find much reference to it on the internet.

The other error I get is "Solutions steps exceed the 15000 maximum". This is partly why I've used such a large increment, to try and get around this.

If anybody can offer some sage wisdom here, I (and my business) would thoroughly appreciate it.

Cheers,

Travis

No one else has this question

Outcomes

Shaun Densberger Jul 10, 2014 2:17 PM

I'd suggest doing a harmonic analysis and determining design changes to minimize the peak response rather than doing a time history analysis to try and simulate the entire event. If you reduce the peak dynamic response, you'll reduce the peak dynamic stresses, thereby increasing your fatigue life.

Actions

Jared Conway Jul 10, 2014 3:39 PM

what dynamic? nonlinear dynamic or linear dynamic modal time history?

what are you thinking the fatigue analysis is going to do for you that you're not going to see in the dynamic analysis?

what do you think causes the damage and failure? let's make sure the study you choose is the right one.

the errors are probalby both related to the length of time you're trying to analyze for.

Actions

Travis Perullo @ Jared Conway on Jul 15, 2014 9:25 AM

Running a modal time history. I couldn't import results from nonlinear dynamic into the fatigue study.

The fatigue anaylsis was recommended by a colleague, as we didn't feel like the dynamic analysis would accurately model the full half hour long dwell.

The damage is caused by the excitation of the Mr/10 rev harmonic on the part. The part is essentially two cocentric cylinders with some circular cuts and slots cut into the cylinders to increase the frangibility of the part. The cracks propogate radially from the ends of these slots.

I figure the 97200 seconds used to try to represent the half hour are why this is failing (it takes well over 12 hours to run the study to failure).

Thanks for any tips you can offer.

Cheers,

Travis

Actions

Nicholas Luyster @ Travis Perullo on Jul 15, 2014 12:28 PM

Hi Travis,

One solution may be to take the high points of stress which you see from the harmonic test and then create hand calculations for the fatigue.

All the best,

Nick Luyster

Gosimulation.com

Online SolidWorks Simulation Training

Actions

Shaun Densberger @ Travis Perullo on Jul 15, 2014 2:14 PM

The only reason to do a time domain over a frequency domain is if the transient solution matters. I can't imagine your system not reaching steady-state conditions within 30 minutes (although you know it better than I do), so a frequency domain analysis should be sufficient.

Run a harmonic analysis and look at the peak stresses, then use the appropriate S-N curve to determine the allowable number of cycles. Within a 30 minutes test at 54 Hz there will be (roughly) your 97,200 cycles.

Actions

Jared Conway @ Travis Perullo on Jul 15, 2014 3:43 PM

The fatigue anaylsis was recommended by a colleague, as we didn't feel like the dynamic analysis would accurately model the full half hour long dwell. > unless you're doing an analysis with nonlinear von mises material in a dynamic analysis, you're not going to see any damage

even with your export to a fatigue, this is just comparing the max stress at a certain location to the SN curve so it may not meet your needs either. like shaun said, run a harmonic and compare with SN curve manually.

Actions

6 Replies

Shaun Densberger Jul 10, 2014 2:17 PM
Mark Correct
Correct Answer
I'd suggest doing a harmonic analysis and determining design changes to minimize the peak response rather than doing a time history analysis to try and simulate the entire event. If you reduce the peak dynamic response, you'll reduce the peak dynamic stresses, thereby increasing your fatigue life.
Like • Show 1 Like1
Actions
Jared Conway Jul 10, 2014 3:39 PM
Mark Correct
Correct Answer
what dynamic? nonlinear dynamic or linear dynamic modal time history?

what are you thinking the fatigue analysis is going to do for you that you're not going to see in the dynamic analysis?

what do you think causes the damage and failure? let's make sure the study you choose is the right one.

the errors are probalby both related to the length of time you're trying to analyze for.
Like • Show 0 Likes0
Actions
- Travis Perullo @ Jared Conway on Jul 15, 2014 9:25 AM
  Mark Correct
  Correct Answer
  Running a modal time history. I couldn't import results from nonlinear dynamic into the fatigue study.
  
  The fatigue anaylsis was recommended by a colleague, as we didn't feel like the dynamic analysis would accurately model the full half hour long dwell.
  
  The damage is caused by the excitation of the Mr/10 rev harmonic on the part. The part is essentially two cocentric cylinders with some circular cuts and slots cut into the cylinders to increase the frangibility of the part. The cracks propogate radially from the ends of these slots.
  
  I figure the 97200 seconds used to try to represent the half hour are why this is failing (it takes well over 12 hours to run the study to failure).
  
  Thanks for any tips you can offer.
  
  Cheers,
  Travis
  Like • Show 0 Likes0
  Actions
  - Nicholas Luyster @ Travis Perullo on Jul 15, 2014 12:28 PM
    Mark Correct
    Correct Answer
    Hi Travis,
    
    One solution may be to take the high points of stress which you see from the harmonic test and then create hand calculations for the fatigue.
    
    All the best,
    
    Nick Luyster
    Gosimulation.com
    Online SolidWorks Simulation Training
    Like • Show 1 Like1
    Actions
  - Shaun Densberger @ Travis Perullo on Jul 15, 2014 2:14 PM
    Mark Correct
    Correct Answer
    The only reason to do a time domain over a frequency domain is if the transient solution matters. I can't imagine your system not reaching steady-state conditions within 30 minutes (although you know it better than I do), so a frequency domain analysis should be sufficient.
    
    Run a harmonic analysis and look at the peak stresses, then use the appropriate S-N curve to determine the allowable number of cycles. Within a 30 minutes test at 54 Hz there will be (roughly) your 97,200 cycles.
    Like • Show 1 Like1
    Actions
  - Jared Conway @ Travis Perullo on Jul 15, 2014 3:43 PM
    Mark Correct
    Correct Answer
    The fatigue anaylsis was recommended by a colleague, as we didn't feel like the dynamic analysis would accurately model the full half hour long dwell. > unless you're doing an analysis with nonlinear von mises material in a dynamic analysis, you're not going to see any damage
    
    even with your export to a fatigue, this is just comparing the max stress at a certain location to the SN curve so it may not meet your needs either. like shaun said, run a harmonic and compare with SN curve manually.
    Like • Show 0 Likes0
    Actions

Cannot recreate failure at 54 Hz Dwell

Outcomes

Related Content

Recommended Content