Different results

[venky40251]

I am wondering why I get different results with Prepromax than code-aster.

Both use same hydrostatic pressure of 0.012 MPa at the bottom of the hopper and 0 MPa at the top.

Both use two bodies (top + hopper) and tied together at the edges. The hopper is fixed from the top edges.

The first picture below if from PrePomax and the second from Code-Aster.

Apart from the difference in deflection, I am not sure anymore which depicts the deflection correctly. Intuitively, Code-Aster seems more realistic.

The I tried two cylinders tied together and fixed from the top with PrePomax and that looks correct

[venky40251]

This is the results from Code-Aster

[venky40251]

this with two cylinders with same boundary conditions

[venky40251]

[FEAnalyst]

You can use the Preview option to visualize the hydrostatic pressure in form of a contour plot and check if it’s applied correctly.

Try with a compound part instead of separate parts connected with a tie constraint. Btw. how are the parts connected in code_aster ?

[venky40251]

Thank you for the quick response. I will try the preview option as well as the compound option. In this case would I still need to create tie constraint for the two parts?

In code-aster I use the liaison_mail which creates a contact between two meshes.

[venky40251]

here is the pressure profile inside the hopper

[venky40251]

[venky40251]

I tried once again in PrePomax without second order for the mesh and get results similar to cod-aster which incidentaly was meshed w/o second order

Now I wonder why so much difference between first order and second order meshes.

[ANYS]

Hi,

Seems a ccx problem not Prepomax. My hopper is symmetric and result is not. Yours seems to have the same issue. (See picture)

[Matej]

@venky40251

Now I wonder why so much difference between first order and second order meshes.

The difference is expected in this case. First-order finite elements can only account for constant stress through their thickness. And in your case, the side surfaces are loaded in bending, which causes a linear stress distribution through the surface thickness. Since you only have one (single) element through the surface thickness your model can not capture bending (linear change in stress through thickness). Second-order elements on the other hand can account for linear stress distribution through their thickness so a single second-order element through thickness is enough to capture bending.

[venky40251]

thank you everyone. Once I made the mesh with 2nd order, both code-aster and PrePomax give almost identical results.

I must say that Prepomax+Calculix workflow is definitely much faster than code-aster.

[arnie]

Typically this would be modeled as surfaces (in midplane) and analysed using shell elements - have you tried this ? It would be interesting to compare.

I 100% agree with your comment regarding the PrepPoMax workflow. Not only is the workflow quicker, it is intuitive and practical. The user interface is way better than Code Aster.

[Matej]

This example also shows that using different approaches (software packages) at the same time can help us get more trustworthy results.