It is widely known that CFX is more stable for Rotating Machinery compared to Fluent. Anyone have any idea how CFX is able to do that ?

Hi,

CFX has a fully coupled approach to the NS equations, but last time I looked at Fluent it used an approach based on SIMPLE which is uncoupled. This makes CFX converge in less iterations and more reliably than Fluent, but at the cost of more memory.

Glenn Horrocks

Thanks Glenn.

Fluent also has coupled solver option also. Still CFD performs better for turbomachinery applications.

Even for imcompressible flows CFX is stable compared to others.

I think it must be something to do with body force term and discretization of convective terms in momentum equation.

Dear Venkatesh,

Fluent coupled solver is explicit and meant for high speed compressible flows.. You may use it near the incompressible limit,but it is not comparable with a fully implicit coupled solver as in CFX..

Also, CFX has only a single solver for all speeds..

Good luck, Opaque..

There are several reasons I can think of:

- *variable* coupled solver, solving for u,v,w,p all at once. Fluent's coupled solver is nowhere near the same technology either. It's virgin and untested really compared to the CFX solver which has been on the market for at least 10 years now. So, saying Fluent also has a coupled solver means nothing really because most of the time I don't think it works.

- the CFX coupled solver *includes* active matrix coupling across interfaces (this is different than u,v,w,p variable coupling), and makes everything really stable. I'm not sure if Fluent does this but I doubt it.

- MFR interface treatments are fully conservative for all interface models. This is not the case for Fluents mixing plane model for example.

- I think Fluent must use full 360 calculations when you do transient sliding mesh. i.e. they have no pitch change models.

Neale

Hi Opaque,

you are correct that Fluent has an coupled Explicit density time marching solver. There also are 2 other coupled solvers: an implicit density time marching solver and a new pressure based coupled solver which is very similar to CFX's coupled solver....ie. solves a coupled (u,v,w,P) set. Density based solvers love high Mach number flows and need to be tweaked to run at low Mach numbers (this is called preconditioning)....so they are not ideal but can run "all-speed"

On the other questions. Fluent has been behind on the other aspects of turbomachinery models/capabilities. ie. mixing plane and non-integer pitchchange interfaces, meshing and post-processing. It seems, however that they are catching up? I don't know if CFX is improving...I think Ansys is more into getting all their tool so they work together in the workbench than new physical models. I was working with Ansys 10 the other day and saw all the same bugs (also known as hidden features ;-)) that I had seen a year ago with CFX 5.7!

The interesting question is who will have the best capabilities for the future...there you will see a lot of people wanting to do transient stuff!

Good question.........

Regards,

Bak_Flow

Hi Neale,

Can you explain about "active matix coupling" or can you give me reference?

And I am not able to understand the defference between Fluent and CFX density based solvers.

Thanks Venkatesh.V