Thin shell for modelling a can 1

[victorroda]

Hello all,

I need to perform a finite element analsysis to determine the deformation of a can under a 2bar internal pressure.

The material of the can is TH520 (tin plate, i've got the strain-stress curves), and the thickness of the plate is 0.18mm - 0.19mm.

I have performed a non-linear analysis with Nastran, and I am obtaining non-satisfactory results, since the nodal displacements are far greater than expected and different from the experimental results. I think the reason is the finite rotations at the nodes, but I am not sure, since I never worked with these low-thickness values. I attach a picture of the mesh.

I have thought of trying Abaqus for this analysis but I don't want to commit the same mistakes, so, which elements would be suitable for this purpose? Any other advise?

Many thanks in advance
Víctor Roda

 

[corus]

The geometry and loading is axisymmetric so use 2D axisymmetric elements. Thick shell elements should suffice but you could also use quad elements to get a better approximation to the non linear stress distribution through the thickness, particularly at the joints. If the walls of the can are long (say greater than 2.5sqrt(rt)) then you could just model part of the can wall and impose a symmetry restraint on the cut edge so that the edge is restrained axially and rotationally.

 

[victorroda]

Dear corus,

thank you very much for your advice, I really appreciate it.

I though about using axisymmetric elements at first, but then the company which manufactures the cans asked me for a buckling-postbuckling analysis, so I decided to have a tridimensional mesh for this study (according to Abaqus explicit user's guide), and then use it for the internal pressure test.

But according to your explanation, I think that two different models will be necessary.

Again, thank you very much for your time.

Kind regards,
Victor Roda

 

[victorroda]

Dear Corus,

I have used an axisymmetric analysis as you suggested me and now the finite element analysis match the experimental results.
Thank you very much

V.