Convert Assembly to Single Solid

If there are gaps between the components how would the FEA give any kind of realistic answer?
TOP

I would be satisfied if there is a method by which these small gaps between parts are filled and then apply Cosmos Express. This would at least get me in the ball park (poor man's version of Cosmos)
Would it be possible to convert an assembly to a single part (using the surfaces only option) and then make this into a solid? Thinking out loud.
I thought Cosmos (full version) does analyize assemblies. Even those with gaps?
Kman

Make ur assembly
then while save it or saves as choose part from the file type, u can choose either of three option i.e. 1 to have surface only, 2 all parts and 3 outer parts and then your assembly (the saved part) will be a single part.
Then u can check for small gaps and fill that gap with solid or surface modeling.
Hope this works for you
Deepak

Just how many components in this assembly? Why not fill the gaps in the assembly?
SW sometimes gets it write and sometimes not when exporting as a part.
Here is a pretty bizarre thought.
1. In your assembly, insert part and extrude a block around the entire assembly. 2. Use the assembly components to cut a cavity into the block 3. Use delete and patch face to remove the slivers 4. Take the resulting cavity and turn it into a solid by repeating step 1 above alt 4. In the part with the cavity make a second block identical to the first as a second body. Then do a combine subtract operation on the solids.
TOP

The assembly has 86 parts of which 28 have clearances between them. That is alot of clean up work. Kman

Gotta say something about common sense engineering here.
How can you expect to get Cosmos to give you an answer when the assembly you have does not describe reality? You can't make a physical assembly with gaps and have any structural integrity until part touch and fasteners or joints take loads.
Reality must prevail. Some rethinking of the overall structure is needed.
Bo

I was thinking that once the assembly was converted to a joined part, then export as a parasolid it would be solved as a single solid part (apparently not). I have run Cosmos Express on the parasolid and received the following message after running the analysis: "Failed Faces (0), Failed Surfaces (0), surface mesh successful, volume mesh failed, try the following -reduce element size (

How are the different parts attached to eachother.
If they are welded, you should after saving the assembly as a 3d-part (not surfaces, or use the guide which TOP described), Extrude the weldments.
Your best opportunity is to make the model as close to the final product.
Also Cosmos (both express and the full version) have some difficulties if you use thin-plate elements on big parts, the minimum thinkness compared to the total size.
Would like some more info on the what it is you are making, and what you need the Cosmos Express calculation for (deflection? stress analyses?).
You should keep in mind that there are a lot of things that can go wrong using Cosmos Express. Do you have any experience using FEM ?

About how the full version of Cosmos works, it builds elements just like the express version, you just have a whole lot of options on how these elements are build, where the Express version comes with a standard set of options.
In both cases the program wont make anything up, it is analysing what you have made. Which means you have to put in bolts or weldments if thats the way its assemblied, if you want the right result.
If you want to "cheat" and do less work, simplify the area you want to analyse. You should have a good idea on where the problems would accur already (if you dont, then the Express version is a "no go" anyways).
So issolate the place where you think the worst stresses / deflection are and constrain it in a proper way getting it as close to reality as possible. This would give you a better result than running the express version on the whole model, cause you are controlling more of the unknows.
Even with the full version, its a lot of work getting the elements build in a proper way and making sure they are connected in the right way also.

The assembly is an equalizing roll clamp and tensioning mechanism. Lift consists of RH-LH core clamp, each of the core clamp sub- assemblies slide independently on a common pair of guide shafts, the roll core cones RH-LH rotate independently.
Interested in both deflection and stress analysis. Areas of interest are the deflection/twist/stress of the guide rods supporting the RH-LH roll clamp sub-assemblies. Twisting of thte RH-LH clamp sub-assembly side plates (load is cantileverd off the side plates).
FEM? none

Purchasing full version is not an option.
What about clearances between bearings, slides etc... (the way it is assembled) How does Cosmos handle component clearances? If I understand you correctly, it does not. That would mean in order to do an analysis one would have to make a new assembly just for this purpose. We are a small company and do not have the resources or time to develop mock assemblies for the purpose of analysis.
I recall a time when we hired an engineer with FEA background via one of our VARS to calculate the deflection of a machine base section for an automotive transfer and assembly line. The deflection results were off by 100%. We paid alot of money for that one simulation and the results were useless. The engineer was provided the actual model and loading conditions.
Simplification: One method being, calculate a beam of equivalent inertia for each RH-LH sub-assembly, attach the beam to the guide rods and run a simulation. It is unfortunate SolidWorks/Cosmos can't close the gaps to make a solid of the assembly. This result would be much more accurate and take considerable less time.
Simplifying and using beam calculations is reasonable for some of the components. Determining the multitude of deflections and stress on the guide shafts is not so straight forward. The load on the roll cones RH-LH is both cantilevered and eccentric. Ideally, the loads are symmetrical on both side.
I would think the results will be less accurate. When you say "controlling more of the unknowns", to me, means making assumptions to simplifying the model. I really would have preferred Cosmos Express (for parts) to simulate an assembly. If SolidWorks JOIN command worked as marketed then this may have been possible. Or maybe, the complexity is still too much for Cosmos to simulate as previously mentioned.
That is disappointing news. The Cosmos (full version) demonstration I witnessed a couple years back didn't mention any shortcommings, grin.

This is a reasonable request. Structural steel fabricators CANNOT have steel on steel (0" gap). I allow 1/16" per side when fitting up to structural members. This works great for guys in the shop, but sucks for me if I want to do any type of analysis with any FEA package. My only solution is to create a variable with the gap distance and factor that in to the length of that part, then I can use configurations to vary that gap number, then turn it on shop drawings parts and assemblies or turn off for FEA. Not clean, but the best I could think of.
Keith

I don't use Cosmos, but I'm sure that Cosmos works the same as Ansys. You have to put in contact pairs at every clearance between components that you expect to touch. The system applies the loads and restraints that you want, and then works out when and how the parts contact one another. This makes the analysis non-linear and usually makes it take much longer to solve, if it solves at all.
If your loads aren't too high, you might be able to get away with building a version of your assembly with the parts rotated and moved into the approximate positions that they will take when you apply your loads. You can then bond the parts together and try an analysis. If you are lucky, make good assumptions, and things behave well, the results won't be too far off. But if any of your assumptions are incorrect or your loads are high enough to cause the parts in contact to deflect appreciably, you will get essentially meaningless results.
You will probably be lucky to get that close.
Jerry Steiger Tripod Data Systems "take the garbage out, dear"

I hear ya. Like many others we download commercial parts from OEM's and distributors whenever possible. And we get what they give us so to speak. Configurations seems to be the best of both realities for now. Thanks everyone for your input!
Kman

Bonding is one of the issues. Join, doesn't work. Creating a part from assembly doesn't work. Manually modifying an assembly as you also mention seems to be the only way this can be accomplished.

In my expirience when it comes to assemblies, remaking it as a part is much faster than adjusting the assembly (making cuts and weldments). I always make another model to make the FE analysis on.
It is a lot of work with assemblies with more than a few parts. But the program aint better than the data you put into it, and there is no such things as Yoda-fields :-)
So you would have to make the weldments / bolt-connections or however it is kept together.
If you save your assembly as a jpeg and send it to me, I might be able to have a look and give some pointers on how to rebuild as a part.
Also I need to know exactly what result you are after. Max Deflection / von Mises stresses in a particular place. If the material stays with the plastic deformation area, if you only allow a certain deflection basicly what you need the result for.

Ronni,
Unfortunately, I am not at liberty to share this information. I do appreciate your offer to help with this problem.
Regards, Kman