[prapanj]

Hi,

I would like to know what the term delta( cell skew factor ) means in the decomposeParDict file. It has a default value of 0.001. I understand cell skew factor as the product of the ratio of the opposite side lengths of a cell, and hence will be 1 for a parallelogram. But how would this affect the decomposition? Does Foam chose cells with the delta value as the processor boundaries? I have been getting a lot of instabilities while running in parallel, code crashes after 6000 iterations etc. But the code runs fine in series. Does this delta affect the stability in any way?

Thank you

Prapanj

[Duncan_vdH]

I have exactly the same question.
Anyone?

Regards,

Duncan

[jms]

Hi!,

did you get any news about delta (cell skew factor)?
I am quite sure that this value affects the stability of the solution.

Thanks,

Greetings.

José

[galap]

*push* Does anyone know an explanation?

[francois]

up
Thanks

[aujamal20]

hello,
Now I am preparing a case to run in parallel. How the subdomain boundaries are defined in 'simple method' by individual solvers......does this delta (cell skew factor) has any influence on this decomposition.

regards,
Jamal

[wyldckat]

Greetings to all!

A test case would be nice, to make it easier to demonstrate the influence of this parameter "delta".

At least in theory (and taking into account that I'm not an expert on this topic), the objective of this parameter is to avoid using the faces of highly skewed cells being used as the processor boundaries, since any solver treatment that could be done on those related cells, would be very difficult to do anything proper, since it would require cells from two or more processor sub-domains. Or if it is able to do anything proper, it'll take longer to run, since it will need several times more feedback from the other processors.

Either way, if you have concerns about these kinds of cells, feel free to increase the value of "delta" to "0.01", to avoid those skewed cells/faces.

In addition, use checkMesh to ascertain the worse scenario of skewed cells/faces your mesh has got.

Best regards,
Bruno

[cfdonline2mohsen]

Thanks Bruno for the useful explanations.
I just want to add that in my experience, using checkMesh for controlling the mesh quality is not always sufficient.

checkMesh -allGeometry

should be used instead.
Foamers can check

checkMesh -help

for more info.