somebody know how to define the constraint between a solid part and a shell? 1

[zhanshenjing]

I am trying to connect a solid cylinder to a shell plate， anyone have some idea how to connect them in Abaqus? Any idea is appreciated. bending moment will be transmitted..

 

[Denial]

I think your specific problem is a particular case of how to transition from brick elements to shell elements. This is a problem I have occasionally wrestled with, in several FE systems. I have not found any totally satisfactory answer, and will be delighted if someone can offer one here.

It is easier to think in terms of the analogous 2-dimensional situation, where you are connecting shells to beams rather than bricks to shells. Consider the diagram below (where I cannot get multiple spaces to work properly so have had to use "." in place of every second space character).[tt]
O–––––O–––––O 2
| . . . . . |
| . . . . . |
O . . . . 1 O——————————————O
| . . . . . |
| . . . . . |
O–––––O–––––O 3
[/tt]
Let the (initial) distance between nodes 1 and 2 be a and that between 1 and 3 be b. If the X direction is horizontal across the screen and Y is vertical up the screen, then you will get the moment transfer you are seeking if your FE program allows you to implement the two constraints
DX2 = -RZ1.a
DX3 = +RZ1.b
where DXn is the displacement of node n in the X direction and RZn is the rotation of node n about the Z axis, etc.
Alternatively you could use the single constraint
RZ1 = (DX3-DX2)/(a+b)
but this is not quite the same thing (and I think I prefer the two-constraint method).

My experience with different FE programs is limited, but I am not aware of any program that allows the automatic generation of these sorts of constraints. To apply the constraints "manually" would be an impossible PIA in the general case, because: (1) you have no ready access to the values of a and b; (2) you might need to define a different user-specified axis system at every location; and (3) in the 3-dimensional case (bricks to shells) the oblique geometry becomes that much harder again.

The limitations with this approach, in addition to those in the preceding paragraph, include:
» It might be invalid in problems involving geometric nonlinearity.
» It will not apply if your shell (or brick) elements have curved boundaries.

A completely different method that is occasionally suggested is to "embed" the beam (or shell) element into the mesh of shell (or brick) elements.[tt]
O–––––O–––––O
| . . . . . |
| . . . . . |
O . . . . . O
| . . . . . |
| . . . . . |
O—————O—————O——————————————O
| . . . . . |
| . . . . . |
O . . . . . O
| . . . . . |
| . . . . . |
O–––––O–––––O[/tt]
However this approach is not without its own difficulties and complications.

 

[IceBreakerSours]

Section 28.6.2 of the Analysis User's Manual (v6.11) says:

Continuum shell elements can be connected directly to first-order continuum solids without any kinematic transition. An appropriate kinematic transition needs to be provided when conventional shell elements are connected to continuum shell elements to correctly transfer the moment/rotation at the reference surface of a conventional shell. Such a transition can be defined with a shell-to-solid coupling constraint or any other kinematic constraint, such as a surface-based coupling constraint, a multi-point constraint, or a linear constraint equation.

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083

 

[rb1957]

i think you should be able to use a rigid beam to join the solid nodes to the shell node.

i thnk the key is to emember that solid nodes only have translational freedoms (ie forces, and not moments) the shell nodes have all six freedoms (forces and moments). so if you connect your shell to a single line of solid elements nodes you're making a pinned connection. if you join to both faces of the solid then it's a moment connection.

 

[Denial]

I'm just thinking out loud here.

If you use rb1957's method without refinement, you are adding superfluous stiffness to the shell element at the transition. Reverting to the shell-beam analog again, and referring to my first diagram above, by running "dummy" beams from 1 to 2 and from 1 to 3 you are making it more difficult for nodes 2 and 3 to experience relative motion in the Y direction (ie to move apart or to move closer together). You need to add some sort of refinement to get around this. Perhaps an axial release in the dummy beams? Perhaps a zero value for their axial stiffness?

 

[zhanshenjing]

Thank you all for replying. I used tie and embeded constrain， both of them produce seriously stress concentration and result none continual stress in the contact area. Rigid connection is not what I want. Finally I used shell to solid coupling that produced nicer results. So Denial, it is proved that the best way to build is shell to solid coupling

 

[rb1957]

but if you don't add beams 2-1 and 1-3 and only attach the shell to 1 you don't transfer moment into the solid ?

and if you "embed" the shell into the solid, how do you avoid double counting ? i meant the solid elements are there representing the shape and now you're adding a shell element (with something like the same thickness property, to make it effective in bending) ?

i don't think rigid beams are double counting the edge stiffness, and if you want to relax the axial stiffness 1-2 (to allow them to "move apart" then you can give it nominal axial stiffness and "plenty" of transverse shear and bending stiffness.

 

[rb1957]

@ the OP, i don't think adding rigid beams 2-1 and 1-3 makes this a "rigid" joint, the stiffness of the joint depends on the stiffness on the shell element, the rigid beams only means the the shell will rotate as much as the bricks are (and the bricks can't tell the shell how much they are rotating directly 'cause they don't have rotation freedoms).

and i don't think there is any "proven ... best way" to do this, i think you have to try different approaches to see what works best in your particular model.

i wonder, could you run a beam along the join and "play" with the torsion stiffness untill the slope is continuous across the join ?