[JuPa]

I fully understand CFX is a 3D solver - to solve in 2D you need to extrude the geometry one cell thick (or if you import a Fluent .msh mesh CFX Pre extrudes it one cell thick automatically).

However why is this such a problem? I hear many people complain that CFX doesn't do 2D. So what?

Am I missing something? Wont a 2.5D simulation in CFX give the same results as a 2D simulation in Fluent?

Can someone explain why people have issues with accepting that CFX does 2D simulations slightly differently?

[ghorrocks]

The big problem with the simple 2D approach used in CFD compared to a "real" 2D solver is that CFX is solving and converging 3 velocity equations, but in 2D flow there is only 2. So you are solving and converging an addition equation. Don't forget that the extra equation is coupled with the other velocity equations and the pressure equation. This results in far longer run time, memory consumption and results file being bigger. Off the top of my head I would estimate a "real" 2D solver to be about 10 times faster than CFX doing an equivalent 2D simulation on an equivalent mesh.

As for your question - won't it give the same results as a "real" 2D solver? Yes, it will. But it will take 10x longer, use more memory and produce bigger results file. It is sometimes harder to converge as well.

There are other less important problems as well - post processing is not as nice as it should be (post processing a single planar field is easy, but CFX requires you to process it as a 3D field. Sure you can just take one of the symmetry planes and you have a 2D field but that should be unnecessary in a 2D simulation, shouldn't it?)

Also some physics such as GGI interfaces, surface tension and some discretisation models are much simpler and more robust in 2D.

I have questioned ANSYS on this many times and the excuse they give is that modern CFD is done on 3D geometries and people don't use 2D simulations any more. I could not disagree more, and the fact that this question about 2D flows in CFX gets asked on the forum so much I think proves it is used frequently.

[evcelica]

Originally Posted by ghorrocks

I have questioned ANSYS on this many times and the excuse they give is that modern CFD is done on 3D geometries and people don't use 2D simulations any more. I could not disagree more, and the fact that this question about 2D flows in CFX gets asked on the forum so much I think proves it is used frequently.

I disagree as well, I probably run more 2D than 3D simulations. And even when something is 3d is nice do get a baseline 2d model done first.