CBUSH or RBE2 for BOLT ?

[minny]

I'm modelling a wing with longitudinal spar and transv ribs

For the flange rib/skin bolted joint I don't know if it's better to use:

1) the CBUSH (PBUSH: K1,K2,K3,K4,K5 high and K6=0 with 3 normal to PSHELL)

2) the RBE2at each fastener location

I have seen same model with solution 1 and same model with solution 2, may you explain the differences? which is more correct?

thanks

 

[thuse]

Bolts are usually torqued enough so that no lateral load is applied to the shaft, with loads normal to the bolt being carred by friction in the materials that are being pressed together. If this is the type of bolt you are using the RBE2 model would be adequate.

If you do decide to put in PBUSHes be careful not to put in an unrealistically large stiffness in any direction. This can cause numerical problems, with the symptom of causing large factor/diagonal ratio warning messages. Also, if the elements that the BUSH connects do not have any rotational stiffness out of plane you can also get numerical problems due to the BUSH "rotating".

 

[40818]

Depends what you want out of the model really, RBE2 will suffice for some instances other ones would require bushes. if you use bushes then you can control teh stiffness of each bolt to represent the correct stiffness based on bolt type, plate material etc, the rigid element wont allow this, hence you would end up with different load distributions between model types.
Neither is more correct, thats why as an engineer you paid to determine which is more accurate for your real life situation.

 

[SWComposites]

Depends on what you are using the model for; if for overall stiffness for deflection, buckling, modes, etc, then RBE's are probably fine (the infinitely stiff RBEs probably don't make a difference in these cases, but you need to carefully evaluate the results). If the model is for load distribution in the joints, local stresses, etc, then RBEs will give the wrong result, while CBUSHes will give a result that is less incorrect but probably still only approximate.

If you really don't have any experience with this type of modeling, you should run some simple joint models using various modeling approaches and compare the results to classical load share analysis results. Then run a simplified version of a wing model using several modeling approaches and compare the results to each other and to results from classical "hand" calculations.