1/2 loading for 1/2 model in ansys?

[YuJie_PV]

hi experts,
I'm working on a finite element analysis (FEA) of a pressure vessel, and I’m using a 1/2 model for the vessel and the nozzle. In this setup, should the forces and moments applied to the nozzle be considered as half of the complete loading, assuming the loading is symmetric? Any advice or insights on how to properly scale the loads in this scenario would be greatly appreciated!

 

[SWComposites]

seems like you would be applying internal pressure loads.
can you show the actual applied loads on the entire vessel.
probably don't scale the loads, but apply that loads that are on the half that is modelled.
but the boundary conditions at the symmetric plane can be complicated. might be easier to model the entire vessel.

 

[IdanPV]

Looks like the PTB-3 example to me.
Why don't you use 2D simplification?

 

[ThomasH]

If everything is symmetrical just apply the full load on the half vessel that you have modelled. But if you use symmetri you wall also automatically ignore certain effects that may be relevant. Like certanin buckling modes, if buckling is relevant.

 

[Denial]

If EVERYTHING is fully axisymmetric, as looks to be the case, why not simplify further and model only a 90° slice? Or, if your FE software allows it, you could follow IdanPV's suggestion and reduce it to a sort-of two dimensional axisymmetric problem. (But note ThomasH's warning about any buckling analyses.)

 

[YuJie_PV]

thanks for all replies.
Yeah, the model is based on the example in ASME PTB-3.
i know either 2D axisymmetric model or 90 degree slice of the model is available.

here is the nozzle loading from my spec. see the bottom row of table:
axial force FA=-12000N;
resultant shear force FR=15000N;
resultant bending moment M2=9840 N.m;
torsional moment MT=9000 N.m.



i just want to learn some general rules which may be applicable to similar scenario of FEA from such example.
please guide me whether 1/2 of the tabulated value shall be applied to be 1/2 model or full value is correct.

any reply appreciated.

 

[Stress_Eng]

As a suggestion, visualize the full vessel and draw a line on it where the cut would be. Then apply each load individually and visualize the deformed shape. You need the line to remain on the cut plane (deformed shape to be a mirror image about the plane). If the line doesn’t stay on the plane, best model the whole vessel.

Edit - You show the vessel (looks physically symmetric); you need to show and consider the methods of assembly attachments and the modelling of their restraint within the FEM (this could introduce non-symmetric reactions).

 

[ThomasH]

If this is for learning i would start with the full model and the full load. The I would continue with half the model and half the load. But this is not a general rule.

To give you a very simple example, that you can try first. Perhaps, just to imagine it will be enough .

Imagine a simply supported beam and two differet load cases. Case 1: Uniform load for the whole span, Case 2: Concentrated load at midspan.

For a model using symmetry you will have no rotation at midspan but it will deform vertically. That is true for both load cases.

However. when you apply the loads they will differ.
Case 1: Full load for the full half beam, then you will get the correct support reaction.
Case 2: Half load at midspan, the "other half" is supported by the now nonexistent beam. The support reaction will be correct.
In both cases the support reaction would be the same for both supports if you modelled the full beam (no symmetri used)

Using symmetri can be very helpful. But many of the "old" reasons for using it, I would say, is no longer applicable. They had to do with computer resources. If you have a large complex model symmetri might be attractive to speed things up. But, can you be sure that symmetri is applicable?

 

[rb1957]

if you're applying moments (and torques) to the model, how are these symmetric ? moment loads are the definition of anti-symmetric loads, so modelling 1/2 the structure is inappropriate.

 

[YuJie_PV]

thanks for all your reply, i think i've got your meaning. thanks