[hulli]

Dear Foamers,

I was wondering, if anybody could help me with that issue.

I am trying to simulate a flow past an obstacle (could be anything a cube an car etc.).
The model itself runs under pimpleFoam, and is a LES model (Wale).

For benchmarking reasons, and to keep the computational time relatively low, I am only focusing at the flow within the domain or the boxes in first a place. In general there are 3 refinement regions so far, ranging from level 0 – 2. I did that using snappyHexMesh (please see the overview figure). At the left side I have a simple velocity inlet with a uniform velocity fixed value (0.6 cm/s), and at the right side of the domain there is an inlet outlet boundary. Front, back, and bottom of the domain are walls/patches whereas the top is a slip boundary.

So here comes my problem:
I always observe strange “clouds / fluctuations or stripes” in the velocity flow field. As far I can tell these seem the result of nut, and become more pronounced in regions of bordering refinement boxes.

Why do I get these strange fluctuations or clouds of nut that affect the flow field?

Why is this mainly happening between regions of bordering refinement boxes and the inlet?

I attached some more figures of Ux and nut as Z slices cut through the domain

… some more background information:
The model uses only U,p and nut.
checkMesh – Mesh OK
The Courant number is below 1
The pressure seams also fine

Thank you so much.

[sturgeon]

Disclaimer: I am very new to OpenFOAM and more or less useless at it. With that in mind, have you tried using non-orthogonal correctors?

Like I said, I have very little experience but it's something someone recommended when I was having similar problems.

See the end of this tutorial:

https://www.openfoam.com/documentati...utorialse3.php

[hulli]

Dear sturgeon,

thank you so much for your suggestion!!!!
I just double checked that, and I have the following settings:
// Outer Loops (Pressure-Momentum Correction)
nOuterCorrectors 3;
// Inner Loops (Pressure Correction)
nCorrectors 2;
nNonOrthogonalCorrectors 1;

I also have the following relexation factors that should do things any better:

relaxationFactors
{
fields
{
p 0.4;
pFinal 1; //< 1 only if alot of nOuterCorrectors (~ 100) are done: Underrelaxation of the final iteration
nuTilda 0.4;
nuTildaFinal 1;
}

equations
{
U 0.6;
UFinal 1; //< 1 only if alot of nOuterCorrectors (~ 100) are done: Underrelaxation of the final iteration
}
}

so I will give that a try ... Thank you

[sturgeon]

You are very welcome, I hope it works! Otherwise I am out of ideas, but I am sure someone else will be able to help.

Cheers
sturgeon

[guin]

check your input from fvSchemes dictionary too

[hulli]

Dear guin,

thank you so much for your suggestion. I am doing that right now. Just in case I also uploaded my fvSolution as well as the fvScheme, if anyone is keen to double check that..

increasing the number of nNonOrthogonalCorrectors did not really help.

Again thank you so much
Best H