FEA of an Assembly

[stu-aspinall]

Hello

Firstly, let me state that I know absolutely nothing about Creo Simulate 4.0, only how to open it and create a mesh within a component.

Our team are discussing how best to go about running fea on Creo assemblies.  These assemblies are made of machined parts, machined from castings.  Our current philosophy for casting to machined is to model the casting, then when we create a machined model, we merge the cast model (using the Get Data > Merge/Inheritance) into the machined model, then add all the machined features, predominantly weld preps etc.

The guys running the FEA, in years gone by, have run into problems using the machined models, as there are invariably gaps between the components, usually filled with welds when the parts are assembled in reality.  If we use the cast models, then invariably they will overlap, which Simulate doesnt like.  We had a student in our team for a while and he came up with a good idea of cutting each cast model, while in the assembled state, using surfaces which he created in the assembly, so that these surfaces had no relationship to any of the component parts.  He then copied and pasted the surfaces into each cast model and solidified them to create a cut in each model, therefore, producing a perfectly mated assembly, which meshed with Creo Simulate, which was his initial prompt that the FEA would run correctly.  Is this something that people use to create perfectly mated up components in assemblies for FEA.

My second topic is, do people generally run their FEA on the least material condition of each component, when in an assembly or the nominal model taken from Creo.  So the components would be opened, and the tolerance analysis > Dimension Boundaries set on all dimensions to be the lower tolerance.  Again, is this something which needs to be considered. 

Also, do people create new models for the FEA, or simply use the actual CAD model.  We have had a suggestion that we also use the Get Data > Merge Inheritance to merge to cast model into a new FEA model.  The link will be there to the cast model so if this is updated, geometry wise, it will reflect through to the FEA model, when regenerated.

I am just trying to get a handle on users ideas for some form of best practices for Creo Simulate....and like I said, I am very green when it comes to Creo Simulate so please keep it simple...lol

Thanks

Stu

[Chris3]

Sounds like you are going to be able to make good use of the new multi-body functionality that will be introduced in Creo 7. It allows for boolean operations within parts.

We use nominal geometry when running simulation results but sometimes use mass growth numbers depending on what we are trying to achieve with the analysis. We rely on safety factors to account for manufacturing tolerances.

Some people use inheritance models for analysis. The advantage is that it is easy to de-feature models and or make changes. The disadvantage is that it is another step and if you are working with an assembly then you have to re-mate everything. Personally I usually either make a copy or use the original when doing an analysis.

[SweetPeasHub]

This is the whole trick of getting a good model.  You will have to develop what works for your technologies.  Sometimes we grow a part some and the cut it exactly by the mating part and use bonded interfaces.  Sometimes we keep the parts clear and select specific small surface regions to weld with rigids between the parts.  Sometimes we develop the mating touching regions to build out that way in manufacturing so we know/gurantee where parts mate.  For example a flat plane could have touch pads in four corners (assuming some decent preload to pull the fourth point in to touch)  If we leave it to chance, we could actually manufacture a teeter-totter because no plane is perfectly flat.  This all depends on how preloaded your contacts are, or if for example they are welded, then just assume connections at the welds, not over the entire plane of "Theoretical contact"  Anyway there are my brief thoughts on the subject.

[stu-aspinall]

Many thanks to all who have replied so far.  I was just trying to get a concensus of opinion regarding analysing the model as nominal/least material condition etc....and how people would contemplate performing FEA on assemblies where the casting models overlap and/or where the machined models would have gaps for welds etc.  Is it better to do the FEA on the cast models assembled together, after performing cuts on the overlapping geometry to make perfectly mated faces.

[GavinBRumble]

Stu,

I know it is tempting to do FEA on design models, and a lot people on this board do so regularly, and a lot more will extol the virtues of doing so.  Roger Toogood, Ph.d., says it this way (in bold letters, too):

   A CAD model is NOT an FEA Model!

I would deviate slightly and say, a design model is not necessarily an FEA model.  The reasons...?

1. Just the very problem you posed in the OP.

2. Boundary Conditions...where actually is (are) the best boundary (ies), and does it (do they) even exist in the design model.

3. Gaps, features, tolerances, features, assumptions, features, slivers, features, materials, features, area of interest, features, setup and troubleshooting, features, solve time, features, reports, features, and UNITS!  Most design models are not created using the units most conducive to FEA.

4. Did I mention $@#*%! features?

Gavin B. Rumble, PE

Creo 6 User 27 years

Mechanica since Rev 14

[SweetPeasHub]

so well said!!!

also. Creo 6 user 24 years. Mechanica structure,vibration,motion. 22years
and still Lovin it! its like playing to me.