Nxy plots for diaphragms

[Anonymous]

Hi,

I am modelling a building with hollow core slabs. Some of them are resting on a large truss. When reporting the Nxy forces in the diaphragms, I get large values in vertical loads cases (see picture attached; load case is only dead loads). This, I think, is an effect of the large deflections of the truss and adjacent tensile strains in the elements.

The slabs are modelled as Deck Slab (one way), with linear release in the connection to the truss. The material is modelled as an isotropic material, with equivalent thickness, adjusted E modulus and G modules in accordance with hollow core slabs. I want the Nxy for exchanging results, so I want to keep the finite elements. Is there something to fix with rigid links? I am not too familiar with rigid links, so please fill me in. 

The results got slightly "better" with Coons mesh for both walls and slabs, but there is still a great discrepancy in results. What are the pros and cons with both elements (Coons vs Delaunay)? Is it someone that has some theory on the subject?

The Nxy plots are for designing the screeding on top the hollow core slabs (if necessary)..

Thanks in advance.

[Rafacascudo · Accepted solution]

If you intend to make the slab " help" the truss on resisting the loads , then I think the model is correct.
But if the slab is just for transfering the load to the truss , then you should release that slab edge horizontal shear.

[Anonymous]

Yes, it is like you say in the latter case; the truss is carrying the slabs and the loads, i.e. the slabs are only loads on the truss. This means that I should release Rx and Ux? Thanks.

Have you anything on the finite element method? Is it so to say that Coons is recommeneded for elements with only in plane forces, while Delaunay is preferable for elements with additional out of plane forces as well? I have watched the webinar, but I can't quite catch the theory behind it and what is making the methods different..  

[Rafacascudo]

Settings for both are different, but I think coons is the only one recomended for curved surfaces.
Yes ,Ux, paralel to the edge