Concrete Shell design with variable thickness

[style480]

Hello,

I've been through the forum and found no answer to my problem.

I am designing an optimal form for a shell, by looking at the deformed shape of a planar plate continually beared on a square under vertical upwards inertial acceleration (equivalent to its weight, but in the opposite direction).

I used SHELL93 elements and it works allright.

Now I want to define a continually changing thickness for my shell (bigger on the sides, and getting always smaller towards the middle). I couldnt find a way to define the thickness as a function of the position: r = f (x,y).

There is still the possibility to define zones and to change the thickness discretely, but I would prefer a continuous change.


I also looked at the topology optimization tools, but i'm not sure there can lead to what I want.


Maybe Ansys can only deal with constant parameters for the real constants set, but is there a way to do it still?


I hope I could make myself clear, and that someone can help me



thanks in advance

++

 

[GregLocock]

Altair Optistruct will do what you want I think, though you may have to use a solid model rather than shells.

Your shell model is a linear taper only, isn't it?

To be honest if ANSYS is anything like NASTRAN world you will have to set up a property card for each particular set of thicknesses, which sounds a bit like your zone concept.

If you really want to stick with shell elements then you could look at OpenFEM for Matlab, and build your own decks, which would allow you to create a property type for each element, so you could make sure that all the thicknesses line up at the corner nodes.


Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[Stringmaker]

You could use the RTHICK command. It will do exactly what you're looking for. You will need to learn a little APDL to use it though.

 

[LeHC]

Thanks Greg to remind me of Optistruct.

I just did a quick check and the result is rather interesting. Attached are screen shots for your comments. Boundary condition: simply supported all sides; load : uniform (not the acceleration). If we allow steep change in thickness, it shows smth like beam and slab structure. If forced to smooth change, the result look like the other file.

Note here that Optistruct show continuous change in thickness through elements, but it is just the visuallization. Inside of Optistruct, the thickness is assumed constant for each element.

 

[LeHC]

Here is the smooth result.

 

[GregLocock]

Red is thick? Blue is thin?




Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[LeHC]

Yes, red is thicker, blue is thinner. Design objective is to minimize compliance (maximize stiffness). I guess if we clamp the sides, the result will be opposite: thicker outside, thinner inside.
I think topology optimization is ideal for this case, if you intend to keep the slab mid-plane flat and change only the thickness.
If you want to design the shape of the mid-plane instead, TOPOGRAPHY optimization is perfect. You can also do simultaneous topology and topography optimization in Optistruct.
Cheers!

LHC

http://cgFEA.com