Distributed Load

[mert1]

I am trying to model a distributed load for Nastran analysis. The shape is a hollow cylinder covered with rubber from inside. I have no idea about the rubber stiffness properties and thats why I ingore the rubber part and would like to use RBE2, RBE3 or some other element to simulate the force acting on the cylinder. This is a normal mode analysis and RBE2 make the piece pretty rigid. Whereas RBE3 needs to be weighted and I am not familiar with it. Is there a better way to go with? Thanks.

 

[johnhors]

Yes there is, DON'T use either a RBE2 or RBE3 both of which are highly artificial and should be avoided when better methods are possible.

What is the force acting on the cylinder? Internal pressure? A distributed load or pressure is generally very simple to apply in any FE pre-processor, what software are you using to create the Nastran deck?

 

[mert1]

This is a bracket that holds(carries) the driveline shaft. Bracket is a circular piece and the shaft passes through the bracket and there is rubber between the shaft and the bracket. I am doing a point mobility analysis. Only the parts on the exterior side of rubber is modeled. So, in the model I have the bracket, the connections to body side etc. no shaft or the rubber part itself is modeled. I want to hit the bracket with an impact hammer and look at the mobilities. Unit force 1N is applied.Initially, I chose one of the grid points on the bracket and applied the load. Later on I applied RBE2 and RBE3 spiders on all the points inside the bracket and hit the center grid point. (For the RBE3, the center point weighting was 0 and all the other nodes were 1). The resulting frequencies matched for RBE3 and one point case(amplitudes do not match). However, RBE2 model was stiff and the frequencies shifted.
I am using Hypermesh for modelling.

 

[johnhors]

Okay, sounds like you have already done it using the RBE3 element, as I understand it, the centre point node (with weighting of zero) will displace as the average of all the other nodes when they have a weighting of one, obviously the weighting can be used as a bias to make some nodes have more influence on the displacement of the centre node over other nodes.

When stiffness is important as in dynamic analysis and stress is of no real concern then an RBE3 element is an acceptable substitute for not modelling the actual structure if it is correctly percieved to have no substantial effect on the outcome of the analysis.

The RBE2 element is still best avoided, as you discovered, since they introduce infinite stiffness into the model. In a static analysis they create rubbish stress results.

 

[mert1]

What would be the alternative then?

 

[mert1]

I have realized something. Actually, I want to hit the middle point and want the other nodes to depend on the center point. In this case, exactly the reverse is happening. So I guess the weighting at the center should be 1 and the rest should be some other value close to zero. However, the center node becomes zero again propabibly because it is the dependent one.