CosmosWorks Modelling question

[ehderube]

Hello,

I'm trying to setup a simple model of a rectangular bolted flange that is under uniform pressure on one side. I wish to examine the resulting displacement of the flange when a bolt is removed under pressure (i.e. will gas escape in this region). The flange is big 10' x 6' x 2" thick with stiffeners, 64 3/4" dia. bolts and 25 psi interior pressure.

My question is what type of restraint do I use for the bolts? I have used radial, axial, and circumferential on cylindrical faces of the bolt holes (all values =0 i.e. fixed) for the bolts and a uniform pressure of 25 psi on the interior face of the flange. When I run the model the von mises stresses are too high at some bolt hole locations. Have I over restrained the bolt holes, if so how do I restrain the model to be more realistic but also stable?

Thanks,
Ed

 

[macPT]

See if thread559-86683 can help you!

Regards

 

[hiyer]

Have you tried using the Bolt Connectors? For this you'll have to include the part that the flange is bolted to. This part can be approximated by a simple plate with a large EX if it is not critical. There are always going to be high stresses at the restrained nodes.

 

[ShaggyPE]

If you are using an older version of Cosmos (as I am, 2004) and don't have the bolted connectors option, you may be able to simulate it this way:

Create split lines on you part for the washer faces of your bolts. Do this on the side of the flange that the bolt heads would be. In your FEA, fix these washer faces. Now apply your pressure to the other side. This should show you your deflection between bolts. Now run a second study where you remove one of the bolt restraints. This should show you what happens when a bolt is removed.

 

[cloche]

Idea of Shaggy is good, but a little bit unprecise if you want to study the exact behaviour of your bolted joint: it doesn't take into account the preload loss due to embedding. But the approximation can be negligible if you first make the analytical study of the "preload triangle" of the bolted joint (use VDI or ISO 90011 norms, it's almost the same). The most exact way to model the bolted joint is by fully modeling it (use contacts...) and then simulate preload by applying a temperature difference (give the plates a zero dilatation coefficient so they won't be affected by that, only the bolts will). This can NOT be done with structure models involving tens of bolted joints, but in order to study one VERY precisely it could do...

Claudio

 

[ShaggyPE]

I don't think ehderube is looking for bolt stresses or hertzian stresses under the bolts. It seems he is looking for the deflection of the plate between the bolts when one bolt is missing.

The only accuracy being lost is the axial deflection of the three remaining bolts under the pressure load. That additional deflection is simple to calculate by hand.
deflection = PL/AE