Undesired cylindrical constraints behaviour - Hydrostatic Analysis

[jeblais]

Hello!

First post here I have to say that I’m fairly new to PrePoMax/CalculiX but I love it! Thanks to the developpers for this free piece of software!

I’m currently testing PrePoMax by comparing analysis’ results with the Nastran In-CAD module for Autodesk Inventor. More specifically, I’m analysing the shell of an open-top water tank when full. To check the robustness of PrePoMax (v2.2.9-dev), I compared the results for the following analysis:

• Full 3D solid analysis - Shell nodes of tank free
• Partial (1 deg revolution) 3D solid analysis - cylindrical constraints
• Full shell analysis - cylindrical constraints on shell
• Partial (1 deg revolution) shell analysis - cylindrical constraints on shell
• 2D axisymmetric of tank of half cross-section. (On a side note, the draftsman did not generate the shell on the positive side of the radial axis, causing nonpositive Jacobian errors, I simply reversed the normal of the plane to solve the error, ignoring the “negative Z” warning.)

For the full shell model analysis, I get non-expected displacements results. I restricted the shell nodes in the revolution direction (theta or T) based on a cylindrical coordinate system but free in the R direction, and it seem the model reacts as if it’s a rectangular coordinate system. Am I doing something wrong? Is it a bug?

Thanks!

[FEAnalyst]

Did you select the local CSYS when defining the BC ? Can you share the .pmx file ?

[jeblais]

Yes I selected the local CSYS for the BC. PMX file is too big to be inserted as attachement, here is a link to the pmx file :

Filebin | hzbe80vr3k6odsvh

Edit: updated hosting site to be less restrictive in download limit.

Also, I should have been more specific, this is what is expected in an axisymmetric manner in the full model, in terms of radial displacement:

Thanks!

[Paul18fr]

Hi,

I played with the model and i can confirm such behaviour when:

• cylindrical coordinates sytem are applied on all nodes
• and the hydrostatic pressure is implemented
  (see modif3 model)

I’va also (obviously) noted that when the pressure is implemented in the gobal system, everything is going correctly.

As the hydrostatic pressure is implemented using CLOAD (pressure recalculated following the 3 components) by PrePoMax:

• it is naturally modified by *TRANSFORM
• as well as by the BCs (component Theta is null here)

I can imagine the equilibrium is modified, so I’m wondering if it the origin of the issue?

Paul

Link here
(available 30 days from01/25/2025)

[Matej]

I can confirm that the hydrostatic pressure is implemented using CLOAD, which is the root of all problems.

[Paul18fr]

Hi,

I can confirm the latest release fix the previous issue using CLOAD (now it is DLOAD); i basically reran the model and displacements are consistent

Thanks Matej

Link here
(validity of the link: up to 05/02/2025)

[Matej]

Great! Thanks for the feedback. May I close this topic now?

[ANYS]

I have compute reaction force in one cone hydrostatically loaded with the new Dload. Compared to analitycal solution deviation is below 0.002%. Structured mesh and medium refinement level.