I would like to know which code can handle natural convection problems real well. Or is this a problem any code can do easily? I am looking at Fluent, Maya(ESC), and CFdesign as three code options. Which of the three is robust in handling this type of analysis?

Hi,

You could also add STAR CD to your list.

regards

David (ICEM CFD Engineering UK)

Hi Stephen,

As far as I know pure (turbulent) natural convection cannot be handled easily by any of the commercial codes running on a pressure velocity coupling algorithm such as SIMPLE, SIMPLER or SIMPLEC. Since natural convection is a temperature gradient driven phenomenon one might need a temperature coupling as well. This has been looked into by Prof. Phil Wood of McMaster University (I will dig up the paper and let you know about the details).

If one uses Multi gridding techniques the problem could be alleviated to some extent. So look for multigridding options in the commercial codes.

Cheers

Mahesh

Hi Stephen. If your interest is only in natural/mixed convection problems, I think that you have to look at some particular features if is or not supported by the software that you are buying. In my point of view, i would choose the software that present most of the following aspects: 1) FVM based 2) If use the natural/mixed convection equations [1] 3) If have temperature dependent fluid properties 4) Easy to geometric modelling 5) If is possible to implement user subroutine

Actually, i am using Ansys to simulate other kind of fluid flow problems, but once i looked in the manual and i didn´t find if the temperature-velocity coupling, Bousinesk assumption, is already implemented in fluid flow equations, i think that there is other coupling method or you have to write your subroutine and use ir as source terms in momentum equations. Among the softwares that i have used, i choose Fluent or Phoenics.

Good luck

Carlos Vilela

[1]Burmeister, "Convective Heat Transfer"

Still, If I have a chance for recommandation of you cfd code, then my recommadation is STAR-CD.

A couple of years ago, I had been used the Fluent. However It was really hard to get the converged solution in natural convection.

turboguy

If the flow is laminar, and if you want ACCURATE results, there is probably no better code than FIDAP, a finite element code within the Fluent family.

You might want to have a look at our FE fully coupled (vel-p-thermal) solver. M-Implict/AcuSolve. It has great capabilities to solve such problems www.mcube.fr