[dseong]

Hi all,

I just want to ask a general question about the CFD validation.

I thought that I needed to do a validation test when I changed my geometry (e.g. add or remove tables/partitions). However, I have seen some research papers using their updated geometry without any additional validation test.

I am wondering how I can confidently say my geometry is experimentally validated. If I add some partitions to my validated geometry and complete the mesh independence test, then can I say that my model is validated without an additional validation test? Are there any criteria/standards?

Thanks!

[aerosayan]

The idea behind mesh independence study is to identify if refining/coarsening your mesh has any effect on your solution.

It doesn't actually prove that your solution is correct. It simply proves that the solution doesn't change when you refine/coarsen a mesh.

Some researchers don't need to do mesh independence study because even on a coarse mesh, their results agree with the experimental/analytical results.

So, they can claim accurately that their solution is correct.

So, before mesh independence study, you should probably find an experimental/analytical results with which you can verify if your solution is correct.

One of my favorite test cases is shockwave over a wedge. It has both an analytical and experimental solution, and to save time, sometimes I run the simulation at a very coarse mesh to just verify if our shock angle is correct in the result, and if the density is changing correctly.

[dseong]

Thanks for your explanation!

Quick additional question,

If I have a simple room geometry which is already completed the experimental validation test and mesh independence test. And now I want to add some tables in this room geom. In this case, do I need to do additional validation experiments with tables? Or am I okay just working on the mesh independence test?

[LuckyTran]

Mesh independence is never a validation.


What if instead of a table, there is an elephant in the room? Your CFD of a room is not representative of a room of elephants and you eventually will need to validate it for elephants.

[dseong]

Originally Posted by LuckyTran

Mesh independence is never a validation.


What if instead of a table, there is an elephant in the room? Your CFD of a room is not representative of a room of elephants and you eventually will need to validate it for elephants.

Thanks! Now, it is very clear to me.