Hello Henry,

To what extent OpenFOam has been rigorously validated and compared with data/other CFD software? Recently I compared skin friction coefficient for a backward facing step case against Fluent's results under identical conditions (i.e. mesh, turbulence model etc). Considerable over prediction has been observed. Let me know if you have any published literature on validation study. I have looked at the following paper: COMPUTERS IN PHYSICS, VOL. 12, NO. 6, NOV/DEC 1998.

Regards
GS

[hjasak]

Heya (not Henry, but still an answer),

The modularity of FOAM allows us to test it part by part, i.e. solvers, mesh handling, discretisation, physical models etc separately. Also, the code has been around for more than a decade and a number of PhD theses and papers contain detailed validation results and experimental comparison for a range of physical models and cases. This includes anything from bread-and-butter fluid flow, free surface, multi-phase, DNS/LES, turbulence modelling to stress analysis and solid-fluid interaction.

Extensive test against leading commercial products have also been poerformed; however the detailed results cannot be put into the public domain due to confidentiality (in short, please feel free to try it out yourself and report on the results).

As for the published material, there is some on my personal web site, especially on the subject of numerics implementation and validation, error estimation and adaptive error control - this for example shows that the code behaves strictly second order because without it I cannot really validate the error estimates. Regarding specific models, I would suggest looking for PhD theses where FOAM was used (have a look at foamcfd.org), some non-linear strss analysis work + extensive LES publications where Henry is a co-author.

Hope this helps,

Hrv

Thanks for the reply. I am kind of stuck, basically I am particular to use/customize OpenFOAM for heat transfer applications. As a sanity check, I am trying to compare results with Fluent for the standard cases. I am not sure if deviation in results is because of numerics or the physical model. Is it possible for you provide some guidelines to systematically check both. I don't mind sharing my results.

Regards
GS

[mattijs]

You could try

Work incompressible.

Eliminate turbulence model. Find a laminar case.

Eliminate the mesh influence. Refine the mesh until the mesh has no influence.

If all this is same then the difference is in the turbulence model and/or wall layer model

Try a low-Re model with a fine enough mesh. If any difference just compare the OpenFOAM implementation ($FOAM_SRC/turbulenceModels/) with that in Fluent.

Go back to high-Re model.

Thanks Mattijs,

Oh, I was under the impression that all this is already done by the developers as OpenFOAM is around for more then a decade. Any way, I will start with flat plat, move to channel, pipe, plane free jet, round free jet, confined co-flow jet, plane wall jet, impinging jet, impinging jet on a pedestal, diffuser, flow over a rib, flow in a sudden pipe expansion, driven cavity and finally backward facing step. All these cases with and without heat transfer.

For heat transfer, how do I include energy equation with simpleFoam?

Regards
GS

[alex_lee]

GS,

I have just set up a group within our division to look into this matter. Perhaps, we could share some of our finding in the upcoming months.

Alex Lee

Thank Alex for your immediate response. Let us work together to rigor and make OpenFoam a robust CFD tool.

Regards
GS

[henry]

I started the FOAM project in 1989/1990 and since then many rigourous testing exercises have been performed by my students at IC and other groups around the world including the backstep case running laminar, RANS, LES and LES with combustion. Some of the validation and research studies have been published in literature, in PhD and MSc theses but by no means all. We still have many of the cases in our archives and could upgrade them to be rerun with OpenFOAM if someone is interested enough to sponsor this work.

It seems your work would progress much faster and be more sucessful if you purchased some support. When setting up OpenCFD and releasing OpenFOAM free and open-source I had hoped that large sucessful organisations such as General Electic, who stand to gain substantialy both technically and financially from OpenFOAM, would sponsor the project or at least pay for the support they need from us and other OpenFOAM developers. Without such income I fear the project is doomed to failure because a stable and reliable support organisation will not be sustainable.

[billy]

Hi,

I sincerely hope OpenFOAM can get the financing it needs, since it is such a great tool for both academia and industry.

[klaus]

Hi Ghanshyam,

i have created a mesh for the NACA 23012-profil were i have some measurements
and solutions from another CFD-tool for validation. But i get strange
results. The total temperature (T+ U*U/2/cp) is not constant in the
domain. Even if the mesh is a bit coarse this should not happen. So i
changed the boundary conditions in a wide (but hopefully still reasonable)
range but this miracle is still there. If you (or others) have the time to
look at it a bit closer i would be pleased to ship it.


PS: See the Topic
Profil naca23012 at about Mach 0.3

Maybe this link works:
http://www.cfd-online.com/cgi-bin/Op...show.cgi?1/868

[eugene]

From what I know this is a discretisation error. We commonly encounter something similar in aerodynamic applications where the total pressure: p+0.5*rho*V2 should be constant but is not.
Refining the mesh should reduce your problem.