Modeling Strategy for High Aspect Ratios

[tfjield]

Hi Guys,

I'm performing either a coupled or uncoupled thermal transfer / CFD analysis.  I have a tube running through my model that carries fluid, and the dimensions of the tube are about 10 micron diameter by about 16 cm long.  The tube is a critical component, as I'm evaluating the effect that the velocity of the fluid has on the heat transfer in the area surrounding the tube.

Is there some way to model the tube other than as a three dimensional solid that would be easier for Simulation to mesh?  As you can imagine, the small diameter of the tube requires a very small mesh, which increases the complexity of the whole model significantly.

A colleague suggested that in other packages he could just draw a line and tell it that it was a fluid tube of a particular dimension, and the software would model it appropriately.  I guess the analogy would be an FEA beam element?  Any think like this in Simulation?

Any help would be appreciated!  Thanks!

[AstroJohnPE]

Wow, I agree that your situation is a high aspect ratio!

Unfortunately, there are no line elements for fluid flow in Simulation Multiphysics. I think a hand-built mesh for the 10 micron tube may be required. Forget for the moment that the mesh of the fluid tube needs to match the mesh of the heat transfer model. Is the geometry of the fluid tube simple enough that you can create the mesh by hand by meshing the cross-section and extruding it ("Draw > Pattern > Move or Copy")? If it can be hand-built, the next step would be to find a method to integrate the external mesh into the thermal model. That is probably not push button but should be doing in some circumstances at least.

With such a small flow passage, is the flow laminar? If so, super-high aspect ratio elements in the fluid analysis may not be too much of a problem.

[tfjield]

Hi John,

The geometry is pretty simple:  it's a fixed diameter tube, but it's not completely straight, it has a couple bends in it.

Using the auto-mesher, I have not been able to get it to create enough small elements in the cross section to incorporate the boundary layers AND show the true parabolic flow.  (Yes, the flow is laminar.)  I can if I use fixed element sizes, as opposed to percentages, but then I get an ungodly high number of elements.  When using percentages, I don't get enough elements in cross section (even at 1%!) but even if I turn off aspect ratio checking it still insists on many small elements along the length.

Any suggestions for things that I might be missing?

As for manually meshing, I haven't tried it because I don't really know how to do it in Simulation.  I suppose I could give it a try someday.

Thanks!