Scared of anything not linear static

[trainguy]

Here's a general question to all you FEA hotshots who perform non-linear analysis, including plastic and crush simulation etc.

How do you validate your results, and feel confident about the correctness of your analysis?

I know that for linear static FEA, I can (4 out of 5 times) back up the analysis with some manual calc. and sleep well at night (if you don't count the crying baby...)

Any comments?

 

[tld23]

trainguy
Test, Test, Test - even if it is a scale physical model or a another simplification of the real structure. Of course you can do simple bounding calculations on many components that act non-linerly, although the bounds may be quite wide. There is no simple single answer, execpt that the validation is too important to 'forget' about!
TERRY

 

[idly123]

well trainguy, anything nonlinear, i would first do a linear analysis first, and as u said may be a hand calc to "validate" llinear analysis results and then go for nonlinear analysis. I belive with the computational complexity involved i should get better than or equal to linear analysis. and obvious we always have the requirement of convergence dicatating the solution accuracy. MAy be with all this one should get a good sleep!!! even after a non linear analysis?!!! Raj

 

[GregLocock]

Trainguy - good question. We build models or prototypes and then spend hours/days/weeks getting the sim and the proto to agree on firstly linear, then simple measures of non linear behaviour, then we test the sims and the protos under increasingly complex behaviours. Some of our legislated tests are too dangerous to be done in the physical world, so once we have a validated sim it is used to sign off these tests.

How do I sleep at night? Well, if the first order non-linearities agree, then in a sensibly designed system the second order interactions won't be too far wrong, and the third order interactions should be negligible. Hoho.



Cheers

Greg Locock

 

[corus]

Validation of non-linear results can be made by using hand calculations that provide an estimate of the solution. The best way I find is to look at the deflections/stresses to see if they make sense and look right. Complex problems generally provide answers that you wouldn't expect but make sense after you have studied them. That's my story anyway.