FEA on assembly yes/no

[NikonF6]

If you are dooing FEA on assembly can you please point some major reasons why not to do it separately on parts instead

when doing assembly, i found that lnumber of parts should be limited to two (e.g. press fit and other contact stress on 2 surfaces)
In all of other case i try to stay away from assembly, the more parts the more errors.
Assembly contains numerous non-natural constraints that multiply error downstream.
On one part, using free-body diagram, i can always adjust and make-up for any expected irregularity, through load or constraints or material metalurgy non-consistancy.
Just one example is a problem with martensite structure only on the surface, which cannot be modeled on FEA, so it has to be compensated manually.

what are your oppinions?

 

[GregLocock]

Consider a system with 3 parts, each joined to the other at different places. Now, with clever use of constraints you may be able to model the behaviour of the whole dynamically, but the chances for error are high. Now do it with a complete car with 2000 spotwelds.

Cheers

Greg Locock


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

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

 

[NikonF6]

Spot welds, no mater number, should be easy to do. They does not need constrains, but parts just can be simple joined into a single part, at spot welds areas. Look at stresses and find the highest stressed spot. Further analysis around that one can made.

But you probably think about that there are so complex assemblies that really need whole things. I have no experience with such thinks as a whole car frame from sheet metal interacting with something else. That should be really tough.

I would rather like to explore something, say, like heavy industry machinery.

 

[GregLocock]

"Spot welds, no mater number, should be easy to do." Each spotweld is modelled with 80 elements.



Cheers

Greg Locock


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

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

 

[inline6]

it depends on the purpose of the analysis. if you are after the global response of a structure you may not need every part on there to do this. it takes experience and judgement though.

there are a few techniques for do spot welds with a single element per spot to gather forces, moments etc but assessment needs to be based on criteria developed for such an approach. many are inhouse and propriety for companies.

ive assemblies of hundreds of parts welded or bolted together and ive done breakaway analysis or sub models. there is no single answer to all this