Clamped shell + Rigid Body


¿What would be the correct way to define this BC.?


It is related to this interesting post by Lucas Bueno.

I can manage to improve the Buckling factor result up to 471.6 but there is no way I can completely remove certain out of plane rotation of the upper shell lip. (See Pict). That is most probably the last detail needed to achieve the exact analytical solution.

How is the rigid body constraint applied here ? To the nodes of the top edge only ? Maybe a layer (or two) of rigid elements would be better.

Very interesting improvement of the results, @Disla. How did you apply the BC?

Hi Lucas,

I was already constraining some additional nodes apart from strictly the ones on the lip but seems it gets better adding a second thin row of elements. (Good point from FEA). A third one seems to much.



My goal is to focus on S8R, KINEMATIC coupling and keep augmenting the complexity of the geometry to see where the laminates are failing if they are.

I still think the issue was on the material definition (Maybe orientation) and local buckling modes. This is the best starting point I can get.

Buckling Load : 473.5 (0.32% error)

Do you have to use rigid body? Ordinary *BOUNDARY constraints don’t work?

HI Victor , How could I do that with *Boundary?.
I need to apply some loads and one Moment to the lip of an structure for further buckling analysis.
I don’t see other option to distribute those loads.
I’m finally using Fixed face on one side and Kinematic Coupling at the loaded side.
Kinematic side shows some disturbaces.

Hi Juan, Victor,

I’m familiar with those BC’s but how do I later apply the Moment?. I need a rigid body, kinematic or Distributed coupling which conflicts.
I could set up the torsion by means of a rotation imposed in the circumferential direction but imposed displacements doesn’t work for buckling.

you have to apply a traction in the loaded edge (y coordinate in NACA’s nomenclature) whose only resultant is a pure torsional moment. In the case of linear elements it’s a uniform distribution of tangential point loads at the nodes of a regular mesh (all elements would have same edge length). Unfortunately CCX doesn’t have this capability already implemented in *DLOAD so *CLOAD cards have to be generated.

Methods to apply torque to a solid body (from Gokhale’s book):

1 Like

I can not understand clearly. Can it be more specific details related to conflict?

I was tested and frequently use coupling feature in CalculiX. Some code changes since versions 2.12 but only minor at input format and still available to the latest.

Another threads discussions at PrePoMax forums i replied. Example input file to download may still available there.

right, these are the practical engineering methods, what I suggested is the numerical version of the mathematical condition so it can be better compared to the analytical analysis (Galerkin) used in NACA TN 1344

¿Could I apply a *Transform CARD to a REF Node ,the set of nodes belonging to the Slave surface or maybe both when defining a Rigid Body or Kinematic Coupling?

I Mean:


¿Is there a way those 1 to 3 DOF refer to a coordinate system which is for example, Cylindrical?

The loading is aligned to global axes, no local axes definition of reference node is required. Only in case of sloped member and load by *Orientation function.

*Transform function is related to edge support since the model is cylinder shell with clamped, a restraint at local coordinate system is required.

So, is that possible?

*Transform function at the edge + that edge being the slave surface of a Rigid or Kinematic Coupling at the same time.?

This is an example of what I mean.

Both shells (quadratic), are under a moment load at the upper lip and clamped on the base.

If the moment is applied through the REF node of a Kinematic coupling, the nodes misaligned and generate peack Stresses.

JuanP proposal to apply the moment as traction generates a uniform deformation pattern but I think would be way harder to set up (if not impossible) in a non-uniform surface like the wing.

Documentation is not clear notified to avoid mixing, so probably it can be use or at least one keyword function is working. Previously i’m only curious about conflict you has been reported.

It"s not a conflict reports, but unexpected behaviour and maybe the limitation itself. It seems the model still need a local rotational edge restraint, so knot can generates making rigid arm. In other way, did *Coupling type *Kinematic can be use as force/moment not displacement/rotation?

This same moment and coupling works perfectly with linear shell elements.
Thit is probably a bug if the moment is not translated into a force for middle through thickness nodes.

This example is no more no less than the moment acting on the stiffeners on the wing.

Probably, cause of linear element does not have mid nodes then make it linearised and produce expectable results. It seems not a bug, need further check on slave definition. Actually, i’m also in learning of expanded element and knot or coupling in CalculiX