Hi,

Just wondering if anyone can help me with a query I have regarding testing for mesh dependency.

I've been experimenting with the global scaling on a model which I have been working with. I understand that with the global scaling option you can "halve" and "double" your mesh size, but that it doesn't change the set spacing options such as edge spacing. I refined my mesh significantly by making the face and edge spacings a lot smaller and then halved and doubled the global scaling. It worked well and I it altered the mesh size accordingly. However, when I use a less refined mesh (it has 680,189 nodes in the final mesh) I get 802,013 nodes when I halve the global scaling and 646,660 nodes when I double the global scaling, which isn't really that far from the original mesh.

I've also found that sometimes when I remesh a model which I have not touched at all, I get a slightly different mesh size than the previous mesh.

I'm a bit puzzled as to these meshing anomalies and wondered if I can't rely on the global scaling, how else I can test mesh independency without going in and literally halving or doubling every value, which doesn't work well for tet elements anyway.

Thanks in advance for any suggestions anyone may have.

Jenny

Hi,

I agree the global scaling factor does not too well in this circumstance. When I do mesh sensitivity checks like this I edit each mesh parameter individually - a pain I know, but it is the only way to get a full mesh with half the mesh edge length.

Glenn Horrocks

Hi Glenn,

Thank you for your reply and sorry for my late one. I have tried to edit each mesh parameter individually but this still doesn't really solve the problem. If you directly halve all the face spacings, min/max edge lengths, expansion rate etc., then there can't be a direct halving or doubling of the mesh which I found. In fact I then had to play around with angle resolutions as the larger elements then caused skewness problems.

In Fluent it is possible to do this quite simply, but I am still a little stuck on how to do a good mesh independence check with CFX.

Any further advice would be much appreciated.

Jenny

Hi,

There is no way I know of to do a direct doubling of the mesh in CFX when you have a tet/prism mesh. You can do a element subdivide in ICEM but then you have to move the interpolated nodes back onto surfaces and curves which never works perfectly for complicated geometries and you have to equivalence the internal nodes.

Therefore I never do a true doubling of the mesh for mesh convergence studies, just approximately double. That is usually good enough for the purpose.

Glenn Horrocks

[Robot]

Dear Sir,

My model is square shape, 2D , simple model .The pink colour represent the reaction between reactant(e.g.glycerol with steam) , with the product (e.g. as hydrogen , carbon monoxide & carbon dioxide). I had key in the reactants & the products already by selecting:

->define
->materials
:name the each reactants & products
-Selecting material type : mixture
-Fluent mixture material : mixture-template
-key in all the reactants & products in properties (mixture species : name)
-key in number of mole for each reactant & product in properties(reaction:finite rate)

All the reactants & products can be obtain from the fluent database.

The red one represent the catalyst. I need to create the catalysts (e.g platinum-titanium oxide , platinum-silicon oxide and platinum-aluminium oxide ) . How should I do it?

CAn I directly change the mesh size in the gambit file ,than convert to mesh file ,to be open in fluent software??

[ghorrocks]

Scaling meshes is easily done in CFX-Pre. But it looks like you are talking about Fluent, so why are you posting on the CFX forum?

[saha2122]

May you Please instruct me abaut Scaling in CFX for mesh study?
Is it automatically?

Regards