Modeling Interactions in Multilayer Shell Structures: An ANSYS/Nastran Approach

[NerseC]

Hello,

I have an issue regarding the analysis of multilayer shell with same material.

The multilayer structure to be analyzed consists of a hollow cylinder with some thickness and sheet metal is to be rolled around it. A simple cross-section demonstration:

_______________________
xxxxxxxxxxxxxxxxxxxxxxxxxx
--------------------------
XXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXX
______________________

--- : contact
X/x : cylinder/sheet metal with same material


My question is, how do I model this in ANSYS/Nastran - which element type or specific analysis (if there is)- so that it will recognize them as different structures instead of one body with total thickness? The purpose of this study is to get frequency response characteristics with different scenarios.

Any help is greatly appreciated.

 

[AlephZero]

You didn't say what the purpose of this multilayer construction is.

If there is any movement between the two layers, that will introduce damping which will have a strong effect on the frequency response. If that is what is intended, look at techniques for modeling constrained layer surface dampers.

If I was modelling this (and I don't know a "standard procedure" for doing it!) I would be concerned about the difference between modeling the curved surface with flat or curved shell elements. The physical difference between wrapping the sheet metal round a circular cylinder or a polygon should be obvious. But even curved shell elements will not give an exact representation of the circular shape ... this seems like a hard problem!

 

[NerseC]

First of all, thank you for your response.

The purpose of this study is to observe the vibration characteristics of the shell structure with different layer thickness/material.

And there shouldn't be any relative movement between the two layers. At least, that is not my intention. The layers will look as if they are glued, except they won't be. FEM modeling, at least, should have them stick to each other like they are pressed.

I'm wondering if this analysis is possible at all using ANSYS / Nastran.

 

[AlephZero]

--- : contact
X/x : cylinder/sheet metal with same material

The purpose of this study is to observe the vibration characteristics of the shell structure with different layer thickness/material.

Either the material is the same or it is different. It can't be both at the same time

And there shouldn't be any relative movement between the two layers. At least, that is not my intention. The layers will look as if they are glued, except they won't be. FEM modeling, at least, should have them stick to each other like they are pressed.

That seems like the standard option for analyzing composite shells, if the interface between the layers can carry shear forces. Basically you define the material properties and thickness of each layer, the FE program calculates the properties of an equivalent single-layered structure, runs the analysis, and then calculates the stresses etc in each layer, if you want to see that level of output.

Your situation is a bit simpler than a typical composite shell where each layer is an anisotropic material and the properties of each layer depend on the direction of the reinforcing fibers, but it should work for isotropic materials.

 

[NerseC]

Either the material is the same or it is different. It can't be both at the same time

That seems like the standard option for analyzing composite shells, if the interface between the layers can carry shear forces. Basically you define the material properties and thickness of each layer, the FE program calculates the properties of an equivalent single-layered structure, runs the analysis, and then calculates the stresses etc in each layer, if you want to see that level of output.

Your situation is a bit simpler than a typical composite shell where each layer is an anisotropic material and the properties of each layer depend on the direction of the reinforcing fibers, but it should work for isotropic materials.

Haha, there will be multiple layers (4-5 etc.), so there could be both same and different material layers in contact :)

If I understood what you'd written correctly, will I not be able to analyze layers of same material in contact as different layers? That is what I would like to know in essence.

 

[AlephZero]

If I understood what you'd written correctly, will I not be able to analyze layers of same material in contact as different layers? That is what I would like to know in essence.

Of course you can do that, for example you can find the stress in the different layers of the same material if you want.

But if the layers are fixed together, I don't see why there is any difference between one thick layer, and two thin layers of the same material.

Think about the simpler problem of a cantilever beam of depth 2d, compared with two beams of depth d stacked on top of each other. If there is no shear force transmitted between the beams, the two beams will be 4 times more flexible than one deep beam. But if a shear force is transmitted because they are clamped together along the whole length of the cantilever, what is the difference between one beam and two?

 

[NerseC]

Ah! Ok.

Here is the concern.

When a sheet metal is rolled around the hollow cylinder, and welded along a line, currently I don't know how the interaction occurs around the whole surface. It's highly nonlinear for sure, but I would like to make some analogy to get more into analytical side.. for example, is it like a bolted joint? glued? or not like any of that? I'm currently doing some experiments on above mentioned joint types and how the FRF characteristics change with thickness/joint variations. In this cylinder/sheet shell interaction, I don't know how it's supposed to be. Of course I can make experiments on it, but I would like to get some reliable information out of this analysis - and make some optimization at the end. That is why I would like to do the FEA, if possible.