I've found this normally occurs when a piece of the model is not properly bonded to the items that are not moving. Static analysis should be that - static. It does not consider bodies that are moving or accelerating. If you have a part with unbalanced forces then it will move unless fixed ie. it is not in equilibrium.
As you note, adding boundary conditions or closing gaps can resolve the problem. If the gap is too large it won't mesh properly.
Sometimes it is hard to find the bit that keeps falling off. I tend to keep fixing bodies until I discover the one that is not bonded correctly. It is a clunky approach. Someone may have better ideas.
Once you discover the body that is moving, ensure that it is bonded correctly.
Most of my problems in the past have been with shells. I believe 2010 gets around some of the previous difficulties.
A screen shot would be helpful.
Here is the screenshot (before analysis and after).
The upper bolts have their shank fully geometrically constrained. The only applied force is a 90 kip load applied normal to the surface of the PTFE inserts. There are contact surfaces between all other pieces.
- The spacer blocks are held in place with the lower bolts and a pretension of 200 ft-lb.
- I have manually inserted body contacts of no penetration with friction for all bolts and nuts
- I have the same global contact condition (just in case I guess)
- Meshing performs fine (no errors or maniuplation needed to get it to mesh first-try with defaults)
- I'm using SW2010 SP0.0 WinXP32Bit
- You can see one nut flys off the assembly for no reason (it has the same contact constraints and BC's as the others - this one arbitrarily leaves the assy with a deflection 10X more than the rest - it's as if it's ignoring the contact constraint).
- I get "Results" after about ten "no equilibrium achieved - save results up until now" error messages and ~12 hours of analysis
Did you model the bolts and nuts or did you use the ones available in the library.
The best way to analyze, would be to avoid the bolts and use a bolt connector instead, this makes the model much simpler and would analyze this model much faster.
The nice thing of using the bolt connectors, is you can specific all the necessary details and you can even very the results buy checking the bolt force.
Okay so you have the answer to your question regarding the error message. Obiviously there is some problem with the bonding of the nut. Have you got he global bond on? Did you define a bond between the nut and bolt. The other thing I've found in the past is that sometimes there are loose nuts hanging around inside the model and you cannot see them. They need to be deleted. Check your feature tree or maybe hide the model to see if this is the case.
I agree with Poovaiah. I'd be using the SW bolt connectors in Simulation. Maybe try a grounded bolt. It is a bit hard to say without seeing what the model is bolted to. Are those bolts really sticking out in the air like that? If not your model may not be accurate.
Posted after your last post. Did you exclude the nuts and bolts from the analysis? Looks like they are included.
Message was edited by: Derek Bishop
I spent some time on the phone with my reseller (very knowledgeable - the first time I've used support like that).
He explained the details of the bolt connector and that having it along withe the nut and bolt parts will lead to inaccurate results. This is the explanation I needed (i.e. what was included with the bolt connector and what wasn't). The bolt connector turns the bolt path into a beam to mimic a bolt. The tight fit bonds the nodes at the surfaces of the bolt hole and the bolt shank.
He didn't have an explanation for why the one net flew off - it's not an extra nut (I know exactly what you're talking about Derek - I've had that before with superfluously copied connector hardware).
I don't have an answer why it flew off like it did - but I suppose it doesn't matter. The correct way to do this is to use the connector.
This is the first time I've used a bolted join with a connector. In years past I used modeled hardware with thermal loads to mimic pre-loads. The "canned" connector makes it a million times simpler.