CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Couplet vs. segregated Methods (http://www.cfd-online.com/Forums/main/4544-couplet-vs-segregated-methods.html)

 Roued March 19, 2002 09:06

Couplet vs. segregated Methods

Hi,

I have a simpel question about the solution process involved in modern CFD methods.

Both commercial and non-commercial CFD codes are often divided about which method to use.

Some perfer couplet method for compressible flows, since they are more robust and faster. This is based on a number of papers descriping

But what about incompressible internal flows ?

Some people claim that there is hardly any difference between couplet and segregated (simple-like, Piso) methods !

Especially in flows involving buoyancy.

If we take a look at commercial code: CFX5 only has a couplet solver, Fluent has both, but as far as I know most peopel uses the segregated solver in Fluent. Star-cd has a segregated solver, also and is going to have a coupled solver in the next version. But they only recomment it for compressibel flow.

Please comment ?

Kind Regards

Roued

 Rami March 20, 2002 04:54

Re: Couplet vs. segregated Methods

Hi Roued,

I have recently read a paper describing a unified approach (using FEM):

S. Mittal, T. Tezduyar, " A Unified Finite Element Formulation for Compressible and Incompressible Flows using Augmented Conservation Variables", Computer Methods in Applied Mechanics and Engineering, 161 (1998), pp 229-243.

Judging from their results, their "blending" approach is successful.

 Neale March 21, 2002 16:25

Re: Couplet vs. segregated Methods

Properly implemented, the coupled approach is ultimately more robust and scalable than segregated. Segregated solvers scale as O(N^2), where as coupled algebraic multigrid solvers scale as O(N). It is quite well known that the coupled solver in Fluent does not work well, and has a limited feature matrix compared with teh segregated solver. I think this is largely a naive implementation issue more than anything else, and not reflective of the behaviour of coupled solvers in general.

The coupled solver in CFX-5 works on both compressible and incompressible flows. There is no option to choose a segregated mode. It's been working fine this way for several years now, and for even longer than that in CFX-TASCflow. I think the fact that Star and Fluent are both attempting to implement a coupled solver is as a result of their realisation that it is the future of solvers in commercial CFD. Very smart people working for Fluent and Star started work on their coupled solvers when they realised how well it was working in CFX-TASCflow. CFX has been using this technology for years now and is simply too far ahead with it for Fluent and Star to catch up in this particular area.

Neale.

 All times are GMT -4. The time now is 09:34.