Shell elements

[Dengr1]

Hello,
I have a problem were a stainless tube (.011 thick) is plated on the inside with copper (.002). I have a heat source generated inside the plated tube and I am trying to see how hot the internal parts are getting? Just a note the stainless pipe has heat sink disks brazed to the outside at a specified pitch which are held at a constant temp and the internal parts consist of another assembly that will be omitted for now to simplify the problem. So to resume I basically have a stainless pipe with copper plating on the inside. I have placed a heat flux on the copper and held the outside surface of the stainless steel at constant temp. I have modeled the assembly as two tubes with the outer tube representing the stainless pipe and a thinner tube representing the copper plating. I then used a bonded contact and meshed the outer tube with bricks and the inner tube (due to its thin thickness) with shell elements meshed using midplane. When I run the analysis no heat is transferred through the copper into the stainless tube. I am not sure why? Did I not fill in a parameter some were for the shell elements? If I use brick elements for the inner tube the mesh is very poor even at fine settings but heat does transfer through (do not trust the results with this mesh). Any help would be great.

Thanks
Steve

 

[GBor]

Well, I have 2 thoughts that come to mind, but you have a modeling error:

Heat will not "jump" across the distance from the mid-plane of the copper over to the inner surface of the stainless steel...that is your biggest problem.

Thought #1
Model the 2-D cross-section of a segment of the piping system as a 2-D axisymmetric model. Basically, cut the pipe longitudinally and take the upper cross-section. Place the heat flux along the inner surface. Doing this, you can make the mesh very fine, get the same or better answer and keep the surface representing the copper in intimate contact with the steel. You should probably model from the center between two heat sinks to the next (1 pitch starting at the mid-point between two heat sinks).

Thought #2
Instead of holding the outer surface of the steel at a constant temperature, provide a convective boundary with ambient temp. If you sent the nodal temperature at the surface of the steel, you want get the cooling from your heat sink and you will actually register higher temperatures than you would measure.

 

[Dengr1]

Hmm, I guess that makes sense since the mesh is midplane, but I thought by making my element the size of the cad model (part .002) would make the surfaces touch. I meshed using the plate and shell mesher option and the heat would not transfer through either. I guess the ultimate question is how then can you set up a heat transfer problem using plate elements wanting heat to transfer threw the parts with shell elements.

Thanks for your thoughts

 

[rybose]

A quick way to see if your parts are connected is to run a modal analysis. It should be pretty clear from the mode shapes if the assembly parts are connected or not.

When meshing thin parts in an assembly I usually have better luck by avoiding using the mid-plane mesher at all costs.

Instead of MP meshing, deactivate surfaces on thin parts and use the plate/shell surface mesher (right click + CAD Mesh Options + Selected for Meshing).

This prevents ALGOR from having make decisions about collapsing surfaces and mainatining connectivity which results in a much more pleasurable meshing experience.

 

[Dengr1]

rybose

I turned off the surfaces for my thin part but I am not doing something correct because the mesher would just crash when. Are you using plate or brick elements? Are you turning off all surfaces or just the ends of the thin part?

Thanks

 

[GBor]

Turn off all but the surface that is in intimate contact with the outer pipe and use plate elements for that surface. You must make certain that all surfaces are in contact for them to transfer the heat.

Can you upload an image of either your model or your geometry. If you are doing what I am picturing in my head, you really should use 2-D axi-symmetry. It would be much faster, better refined mesh, and produce at least as accurate results.

 

[Dengr1]

I will try the 2-d approach but I do need the 3D model to run because I need to have it in my total assembly which includes the internal components generating the heat and the external heat sinks.

Thanks GBor and rybose for your time and comments. I have found that there are not too many forums to go to with respect to algor. It is nice to get feedback from other users.

I would like to upload an image of my model but unfortunately I have been told not to. Sorry about that.

Thanks

 

[rybose]

Sorry I wasn't clear in my original posts.

Like GBor said, turn off all of the surfaces *except* the ones you'd like to mesh with plates/shells.

If you try an mesh a part with no surfaces "selected for meshing", the mesher will give you an error just like you said.

 

[rybose]

Sorry I wasn't clear in my original post.

Like GBor said, turn off all of the surfaces *except* the ones you'd like to mesh with plates/shells.

If you try to mesh a part with no surfaces "selected for meshing", the mesher will give you an error just like you said.

 

[rybose]

Sorry I wasn't clear in my original post.

Like GBor said, turn off all of the surfaces except the ones you'd like to mesh with plates/shells.

If you try to mesh a part with no surfaces "selected for meshing", the mesher will give you an error just like you said.