Interpreting FEA results

[arm100]

Guys, I am performing design analysis using ANSI/ALI which is mostly based on hand calculations. Obviously, my stress analysis results are going to be higher than those obtained by hand calculations. Is there a fudge factor I can use to address this discrepancy?

Thanks.

 

[GBor]

Obviously, my stress analysis results are going to be higher than those obtained by hand calculations

No, actually it isn't obvious. Can you tell us a little bit about the geometry that you are working with? Do you have geometric discontinuities?

I recommend you do some design validation problems to make sure you are using the software properly...some simple geometries from Roark's or your college textbooks.

 

[corus]

I don't know the software but hand calculations are to give you an indication or a check on your FE results. The FE results will, or should be, more accurate. There are no fudge factors.

corus

 

[mechengdude]

I concur with the other posts that your FEA results shouldn't neccesarly be higher than hand calculations, just more accurate on complex geometry.

I'm wondering if you are referring to results associated with restraining bolt holes or other similar geometry which can result in singularities and values that are artificially high.

 

[arm100]

The weldment is complex and there are stress raisers due to welds, etc. But for this type of complex geometry, traditional Strength of Materials solutions are for checking FEA results, as noted in one of the posts. I have been doing FEA since the 70's so I am not a novice, but the code used in this case is very restrictive because of uncertainties in the hand calculations. So the question is: How do you use a design code that was meant to address hand calculations when you use FEA instead?

 

[GBor]

Is there a specific code that you are dealing with? ASME Boiler and Pressure Vessel? A specific piping code?

 

[crisb]

Its an interesting question to which I dont think there is an easy answer.
Occasionally one gets stuck with old permissible stress codes when trying to use FEA. At its most ridiculous someone will compare peak unsmoothed FE stresses based on stresses at integration points interpolated to nodes, to the allowable stress in the code.
If you cant ignore the permisible stress code or change to a limit state code, then it seems you could be destined for an uneconomical solution.

 

[mechengdude]

O.K. I understand your question better now. I have used specific hand calculations defined by code in the past for pressure vessels but never applied an FEA solution to a code written for hand calculations.

I have also dealt with independent third party inspectors that verified the stress numbers and the 100% inspection of the vessels. I would contact the inspector if this fits your case or the governing body that wrote the specification. The governing body should be able to aid you.

Good luck.

 

[corus]

Design codes are written for hand calculations in the main and as such the safety factors for the assumptions you've made will be generous. You should find though that pressure vessel codes, for instance, will have a separate section for 'Design by Analyis' which are aimed at FE results and are more specific in their stress classification. Peak stresses that you'll find in your FE model from stress concentrations will be treated as causing fatigue damage, and stresses at structural discontinuities will be treated as secondary. I'd refer to their guidelines, whatever the structure is, for assessing these 'anomalies' and use your main design code for assessing the nominal stresses away from such features.

corus

 

[Struct71]

I see a parallel with the following thread.

http://www.eng-tips.com/viewthread.cfm?qid=211759

There is a black hole in the codes when FEA is used,
and that is why you often find statements like "The maximum stress is local therefore allowable".



And I agree that FEA usually gives more accurate but also higher stresses.

What we need as engineers is a solid guideline or code that defines a) what stress to use - ie which equivalent stress or which axial stress, b) what is maximum allowable, c) how big and large can peak stresses be.

So far I have not found answers to above issues.