[itchy]

Hi all,

I've done some LES simulations of pipeflow with dynOneEddy-Model and compare my results with oneEddy-Model and WALE model. I get some strange results for ksgs -> nuSgs.
first some informations:
1) problem description
- pipe diameter 0.04m
- length of the pipe 4*D (cyclic Boundary conditions)
- fluid water nu=1.081e-06
- incompressible flow Re=4900, Retau=340
- Grid: structured O-Grid
r+ wall = 2,5
r+ mean = 7.5
z+ = 35
eta+ = 10
I have also used finer grid and I have the same problem (problem see below)
2) Boundary conditions:

---k , p , nuSgs---
boundaryField
{
inlet
{
type cyclic;
}
outlet
{
type cyclic;
}
pipewall
{
type zeroGradient;
}
}

---U ---
boundaryField
{
inlet
{
type cyclic;
}
outlet
{
type cyclic;
}
pipewall
{
type fixedValue;
value uniform (0 0 0);
}
}

3) Schemes/Solution/Options
- I used pressureGradientExplicitSource to get the cyclic flow.
- all schemes Gauss linear or Gauss linear corrected
- ddtSchemes: backward
- pimple with residual control U,p<10^-5, k<10^-7
- initalize the U-field fully turbulent by mapping data from other simulation
- fieldAverage 200D

The velocity profiles in u+ y+ looks ok for all models (dynOneEddy is ok but the others are better). But in dynOneEddy-model I get strange behavour for nuSgsmean and kSgsmean.
--------- Here my results for nuSgsmean over 200D----------
1. WALE:

SmagNuSgsmean.jpg

looks good. At the wall, nuSgs decrease because of WALE model. NuSgs is smooth ober the domain

2. OneEddy

OneNuSgsmean.jpg

results looks reasonable, because I do not use van Driest damping at the wall and it is not a dynamic procedure implemented. But nuSgs is smooth.

3. dynOneEddy

dynOneNuSgsmean.jpg

looks not good!! Why these two stripes???

------ Here my results for kSgsmean-------------
1. OneEddy

OnekSgsmean.jpg

looks reasonable, because no dynamic procedure implemented.

2. dynOneEddy

dynOnekSgsmean.jpg

puuh, What's going on at the wall?? What about these two stripes again??

In dynOneEddy-model the results for nuSgsmean and kSgsmean looks strange. First I thought it is the grid. But even with finer grid I get the same results. For Smagorinsky, dynamicSmagorinsky, WALE, OneEddy-model I get with this mesh, bc etc. good results. Whats wrong with dynOneEddy-model???
Is there a fault in the implementation of the dynamic procedure in dynOneEddy-model??

Somebody an idea???

kind regards
Florian

[itchy]

Hi,

is there somebody with experience with dynOneEddy-Model????

kind regards
Florian

[zhaoshiyu]

Originally Posted by itchy

Hi,

is there somebody with experience with dynOneEddy-Model????

kind regards
Florian

Did you finish this problem
I also have some problems using this scheme