Global model to local model (Submodelling technique)

[Aero9397]

I would like to have some information, on how to analyze the local models by taking the loads from global model.

As per my knowledge there are various methods like
Reduced integration method.
Stitching the local model to global model
Getting free body loads from global model and applying on local model
getting enforced displacement at boundary and applying on local model.

Please let me know if you have any information on above by mentioning the merits and demerits of each method.

Thanks

 

[Aero9397]

Anyone has more info on my query,just checking?

 

[RPstress]

I have never used reduced integration for this and so cannot comment.

Embedding a detailed model in a coarser global one (i.e., just refining the mesh in an area of interest) is a good technique if the transition from coarse to detailed is done correctly. Most systems include interpolation elements to make this easier.

A variant on the above is the use of superelements, where the model other than the sub-model is reduced to a stiffness matrix.

Taking the nodal boundary loads from the global can work. A method of interpolation from the few global nodes to the many peripheral sub-model nodes is needed, or the sub-model can be made bigger than necessary and match, say, every 10th sub-model node to a global one, although this is not usually done; some sort of interpolation is the norm. Again, interpolation elements can be used (e.g., a Nastran RBE3).

Enforced displacements need similar interpolation considerations to loads, but the stiffness of the sub-model becomes critical. Even a small difference can make a big change to the forces generated. This method is probably more appropriate to a sub-model of a structure with a discrete connections to a larger part, e.g. a flap on a wing (especially if the flap is like the outboard one on the DC-10, with more than two flap tracks so that wing deformation also bends the flap).