Mesh Size CAD geometry and Manual Meshing

[Boghi1990]

Hi GBor(Mechanical),

Is there any basic guideline which can give an idea on what would be the appropiate mesh size when one starts meshing a CAD model imported into Algor? The same question if the mesh is created manually in Algor FEA editor ?

Is the mesh size dependent of the actual dimensions of the part intended to be meshed ?

I am wondering if there is a general rule that I could follow when dealing with intial meshing of the model.

Attached is a model(insulation 3", pipe 30" OD and fluid)created in a CAD package)that is to be used in Thermal Transient analysis. The length is 1 meter.

I am looking forward to hearing from you.

Kind Regards,

ONEPOINT

 

[GBor]

Considering the number of quality users on this site, I'm honored to be "singled out" for a specific question.

This is a GREAT question and one with very little answer. Thermal analysis generally doesn't require the same mesh as structural unless you plan on doing thermal stress, but whether you are auto-meshing or manually meshing, one this that is rarely done and should ALWAYS be done is a mesh sensitivity analysis.

Basically, you should plan on running each analysis model at least twice: once to establish a baseline, once to refine the mesh and see if you answer changes significantly.

For the particular model, is there a reason why it needs to be 3-D? Are you attempting to account for thermal exchange and the ends? If so, you can still use a 1/4 symmetry model and save yourself a great deal of calculation time.

As for mesh size, the auto-mesher takes the shortest edge, divides it into a certain number of segments and uses the length of one of these segments as the baseline element size. If it is unable to mesh at that size, then it drops down up to 6 times decreasing the attempted size by a factor that you can set in the dialog box. You can switch the auto-mesher to try an absolute mesh size to begin.

After performing many analyses, you will eventually get pretty comfortable with an initial mesh size. You shouldn't abandon the practice of mesh sensitivity analysis (I say somewhat hypocritically, since I don't ALWAYS do this...but I should).

This has been a long drawn out explanation to say:

You will have to use some engineering judgment. For stress, you will want a more refined mesh in areas where you expect stress concentrations. For thermal, you usually want a very even mesh.

Use symmetry where you can to keep run times down.

 

[Boghi1990]

Hi GBor,

Thank you for your answers. I am adding here just a few more comments and questions. That model was used for a transient thermal analysis to determine the cooling time of the bulk/mean temperature of the stationary fluid. So given this situation, do you think that a 3D model was not appropiate to get the desired results ?

Could I use symmetry if the model is a large storage tank filled with fluid(natural convection),where the tank is exposed to a windy environment at a temp of -30 C? The same intended result as for the pipe model is required in this case. Looking for the cooling time required for the bulk temperature of the fluid in the tank to reach a certain temperature. So could I use symmetry or do I need to model the whole tank?

Another case would be if the same tank would be analysed for linear stress to get the Von Mises stress in the area where the vertical walls connects with the base plate? Could I use symmetry for this structural analysis?

Looking forward to hearing from you.

Regards,
OnePoint